Condenser Water Systems, Air Entrainment, and Pump Cavitation

This post is the result of a discussion I was having as the result of a comment made on a previous post about commissioning a condenser water system.  The initial question was to ask if the solution I had suggested to the problem that was the focus of the post – basically air venting problem – had worked.  The answer to that is “yes”.   The manual air vents were replaced with automatic air vents and, to the best of my knowledge the problem was resolved.


You can find some more details and close-up photos of the air vents in my reply comment in the original post if you are interested.

The person I was corresponding with thanked me and mentioned that they thought they had a similar issue in a system they were working with and that one of the symptoms was that there was a crackling noise in the pump suction when it was running at full speed/design conditions.  I started to reply because there are a number of things that can cause a noise like they were describing.  But as the reply grew in length, I realized it would probably make a good blog post or two. touching on a number of condenser water system topics, so here we are.

The links below will take you to the various topics that I ended up discussing.  The end of each section has a “Back to Contents” link that will bring you back here.

Air Entrainment

Air can definitely cause a noise like the one that was described in the comment. Condenser water pumps that are located very close to the tower they serve are particularly prone to this if the tower outlet is undersized or the flow rate is higher than the tower is designed to accommodate because a vortex forms in the tower basin at the suction connection and entrains air into the piping leading to the pump.

This exciting video clip is one I recently shot that shows a vortex starting to form at the connection to a sump in a cooling tower.

Back to Contents


Having said all of that, sometimes a crackling or banging sound on the suction side of a pump is cavitation, which is related to the velocity at the inlet of the pump and occurs because of the reduction in static pressure associated with the increase in velocity that occurs as the water accelerates into the eye of the impeller.

In general terms, the conversion of static pressure to velocity pressure that occurs as the water accelerates through the smaller cross-sectional area (relative to the pipe) of the pump impeller eye and impeller channels can cause the static pressure at those locations to approach and drop below the vapor pressure of the fluid that is being pumped.  The is especially true for condenser water systems where the pumps are only slightly below the elevation of the tower basin (meaning not much static pressure available to keep things positive at the pump suction) and especially if there are a lot of things in the pipe between the tower and the pump that can drop the pressure (like a plugged strainer).

The image below is from a pump manual that I have – The Durco Pump Engineering Manual – that shows the pressure gradient as a fluid approaches and moves through a pump impeller.  As a frame of reference, the upper image is a cross-section through the pipe and pump with the pipe to the left and the impeller, volute, and pump discharge just to the right of center.   The chart below the image shows the pressure gradient through the cross section;  the letters on the x axis of the pressure gradient diagram correspond to the locations referenced by those letters on the pump cross section.

Impeller pressure profile from Durco Pump Engineering Manual

If the absolute pressure at any point gets below the vapor pressure of the fluid being pumped, then the fluid will “flash”; i.e. change phase from liquid to vapor.

There is a tremendous volume change associated with a phase conversion from liquid to vapor and vice versa. For instance, if you convert a cubic inch of liquid water to water vapor at atmospheric pressure, the volume changes by a factor of about 1,600; i.e. 1 cubic inch of liquid water becomes 1,600 cubic inches of water vapor. When that happens in a confined area, like a pump impeller for instance, there is a force generated, which can make noise and also cause damage.

Of course, since pumps are designed to pump liquid, not vapor, as soon as the liquid flashes, the flow produced by the pump is reduced significantly. And, since it was the pressure drop due to flow that caused the low pressure that triggered the flashing, as soon as the flow drops, the pressure goes back up, and the vapor implodes back to a liquid, so another huge volume change and more noise and force.

When cavitation is triggered, the phase change phenomenon I described above is happening over and over again and that is what is causing the noise and what also can cause damage to the pump.

I should also note that while cavitation is most common in the pumps in our systems, it can also happen at the control valves, especially in situations where the valve imparts a large pressure drop to the fluid going through it.   The results are similar;

  • There is a lot of noise, and
  • The valve can be damaged, and
  • Since the valve is probably not as firmly anchored as a pump, the valve and pipe can actually start moving around.

The pipe movement can alarming; inches, not just fractions of an inch, and definitely will catch your attention.  At one point, Fisher Controls published a bulletin on the topic, which is how I learned about it.  If you are interested, I have a copy of it on my website and you can download it from there.

Back to Contents

Compound Gauges

The potential for sub-atmospheric pressures at the pump suction flange in situations like the one I just described and the ensuing potential for cavitation make it a good application for a compound gauge.  Most pressure gauges have the low end of their range at 0 psig.  The “g” in psig means “gauge” and indicates that the pressure indication is relative to atmospheric pressure.

But it is also possible to reference a gauge pressure reading to an absolute vacuum and those readings are termed psia, with the “a” indicating the reference is a vacuum.   For example, a reading of 0 psig at sea level is the same as a reading of 14.7 psia or 30 inches of mercury.

A compound gauge is a gauge that references a vacuum as the low end of its range.  Typically, the range will start at 30 in.hg, which stands for 30 inches of mercury vacuum (Hg is the chemical symbol for mercury if you look on a periodic table of the elements). The range will typically convert to psi at when it reaches atmospheric pressure, where it will have a zero, to indicate 0 psig.   From there, the range increases to full scale with the numbers representing psig.  Here is an example of a compound gauge along with a picture of what the internal workings of a typical gauge look like right below it.

DSCF5194 Gauge guts 01

This particular gauge has a handy feature called a “Maximum Indication Pointer”.  The red pointer is not directly connected to the internal mechanism.  Rather, it is arranged so that the black pointer (which is connected to the internal mechanism) can push it clockwise.   By turning the little brass knurled nut you can see sticking out of the center of the gauge, you can manually turn the red pointer counter clockwise so that it rests against the black pointer.

So, if you have the gauge connected to a manifold that allows you to take a number of readings using one gauge, like this ….

Preheat pump gauge piping

… then if you take the high reading first – say the pump discharge pressure – the red pointer will be pushed to that indication.  If you then take the suction pressure reading, the high reading is “retained” by the red pointer while the lower pressure reading is shown by the black pointer, which allows you to directly read the difference in pressure off of the gauge, which is usually the number you are actually interested in.

I should mention that one benefit of using one gauge to take multiple readings on a given piece of equipment is that it automatically will eliminate gauge accuracy as a source of error since you are subtracting readings taken with the same gauge.  And, you also don’t have to account for the impact of elevation differences on the gauge, which you need to do if you are using two gauges that are at different elevations to measure a pressure drop across something.

Back to Contents

Impeller Damage from Cavitation

Aside from making a pump under-perform, cavitation can do significant damage to the impeller.  Below are a couple of pictures that Jeff Mahin – a student in one of the classes I teach – shared with me. They were taken through an open suction diffuser associated with a condenser water pump.

Cavitated Impeller on Middle CW Pump Cavitated Impeller on Middle CW Pump Zoomed

As a part of a field exercise, his project team had run a pump test and the numbers were not adding up.  So we opened up the suction diffuser to see if there were any clues as to why. The second image is a close-up from the image to the left that zooms in on the impeller. In it, you can clearly see how the impeller has been chewed up by cavitation which is likely the reason the pump was not performing per the published curve associated with it.

The next string of pictures where shared with me by another student – Jay Cmiel – who has been trying to understand issues he has been having with a domestic water booster pump. At this point, he believes there are a number of things going on, including system configuration issues and possibly metallurgical issues. But he also suspects that cavitation is a major player behind the failures he is seeing.

These first two pictures illustrate what the impeller in one of his domestic booster pumps looked like back in June of 2016 after it was installed to replace a damaged impeller. The new and old impellers are shown side by side in the second picture.

Damaged Impeller 05 Damaged Impeller 02

This picture is what the impeller looked like in September of 2016 when they opened up the pump to replace a seal that had failed.

Damaged Impeller 06

These last two pictures are what the impeller looked like a couple of weeks ago (late April/Early May 2017) when they had to replace it again.

IMG_5214 IMG_5216 IMG_5217

So clearly, the message from the field is that cavitation is a thing to be avoided!

Back to Contents

Net Positive Suction Head (NPSH)

One way to understand if the potential exists for cavitation is to do a Net Positive Suction Head Available (NPSHa) calculation.   The basic idea is to do a calculation that tells you what the absolute pressure will be at the inlet flange to the pump in question at its design flow rate given the configuration of your system and its warmest operating temperature.   That number is is the Net Positive Suction Head Available or NPSHa.

You then compare that to the Net Positive Suction Head Required a.k.a. NPSHr for your pump.   That information shows up as a line on the pump curve.  For example, in the curve below, the design point for the pump is 1,000 gpm at 45 ft.w.c..  At that condition, it will be about 83.5% efficient, use about 13.25 bhp, and will require about 11.2 ft..w.c. of net positive suction head (or more if you want to be on the safe side) to avoid cavitation.


Notice how the NPSHr curve is read off of a secondary axis to the right and that it tends to rise as the flow increases because for a given cross-sectional area, an increase in flow will cause an increase in velocity.  And since the velocity pressure is  function of the square of the flow, the curve is not linear and really starts to take off as the flow increases.

The latter item is an important consideration in the field because it means that a pump that is just fine at it’s design condition can cavitate a lot if it runs out its curve.  That phenomenon is quite common in systems where large pumps are piped in parallel and serve common headers.  Consider the system illustrated below.


The three relatively large pumps (one is a back-up pump) serving the system are piped in parallel and deliver water to and from the chillers they serve via a fairly long piping run.  If both chillers are running, then the pumps need to deliver the design flow for the chiller at the head generated by that flow in the long piping run;  in this particular case, the total flow requirement was 2,400 gpm and the head requirement was 65 ft.w.c..

But if only one chiller is running, the required flow drops in half and as a result, due to the square law, the head required at the lower flow rate for the common piping that the pumps share is only one quarter of what it was (one half squared or half of a half).  As a result, in this particular case, when one pump was running, it ran out its curve and delivered significantly more flow that was required by a single chiller when it was in operation;  about 1,920 gpm at 47.5 ft.w.c. when only 1,200 gpm was required by the chiller.


That turned into an energy conservation opportunity that was worth anywhere from $1,100 to $15,000 a year depending on how you captured it.  If you want to know more about that, there is a PowerPoint case study on my website that you can look at which includes the cost/benefit metrics for the various options that were possible for capturing the savings.

My point in bringing this up here is that when system dynamics cause a pump operating point to shift out its curve significantly, the operating point on the NPSHr line can shift from a condition that was satisfied by the NPSHa to a condition where the NPSAa is not sufficient to prevent cavitation.   If you don’t fully comprehend the dynamic behind what is going on, then the cavitation problem can appear to mysteriously come and go.

For the system in the example above, this turned out not to be an issue.   But I have seen it be an issue on other systems with similar configurations and similar dynamics, which is why I bring it up.

The first time you do a NPSHa calculation, it can seem a little intimidating because you are working with absolute pressures and things like that;  it kind of seems like you are on a different planet or something. But Bell and Gossett publishes a handy little nomograph (chart that lets you graphically determine a solution) that makes it pretty easy.

B&G NPSH Nomograph

Us old folks have it in our Bell and Gossett Engineering Design Manual.

Bell and Gossett Manual

But you can now download electronic copies of the various bulletins that are in the manual from the Bell and Gossett website.  The nomograph shows up in the one on cooling tower pumping and piping, which also includes a lot of other very useful, practical information about cooling tower and condenser water system design and operation and how to apply pumps in them.

Back to Contents

So, I guess that ended up being a pretty long discussion for something that started out as answering a question about air venting.  But hopefully, the information is useful;  if it is, you can thank Kam for asking the question.


David Sellers
Senior Engineer – Facility Dynamics Engineering

This entry was posted in Chillers and Chilled Water Systems, Operations and Maintenance, Retrocommissioning Findings, System Diagrams and tagged , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s