Parallel Pumps; An Indicator of a Retrocommissioning Opportunity

If you study the system diagram in my previous post that described the heating hot water system that I am discussing, you will notice that the HVAC pumps are to some extent, in
parallel with the Process pumps.  I have reproduced the system diagram below with both sets of pumps highlighted in yellow for your reference in the following discussion.

HW-Higlighted-Large

I say “to some extent in parallel” because while both pumps share the same suction header (upper green highlight), they do not share a common discharge point until after they have pumped through the loads they serve and connect to the return header (lower green highlight).  If they were directly in parallel both the pump suction and pump discharge would have a common point of connection prior to serving the loads.

But, you will notice that there is a balance valve on the discharge side of the process pumps (blue highlight).  This valve is probably there to allow the system to continue to circulate water even if the process loads area closed off.  Maintaining some circulation is desirable in that it will prevent pump heat1 from building up in the pump body and causing cavitation.  It also helps maintain the water chemistry and water quality.  So spending a little energy on recirculation is probably not a bad idea.

The process pumps are a different type of pump from the HVAC pumps (in-line vs. base mounted end suction) and likely have a different pump curve.  Based on past experience, seeing pumps with potentially different characteristics in parallel is a “red flag” of a potential interaction that may be different from what you might expect and cause unanticipated operating results.

Specifically, if one of the pumps has a pump curve that is capable of generating pressures in excess of the shut off head of the other pump, then in some operating modes, the pump with the higher pressure capability can stop the other pump from producing flow, just as if you had closed a valve.  This can create mysterious loss of flow  problems until you realize what is going on and make a change to correct it.

To illustrate this problem, consider the following pump curves, which illustrate a situation we have set up at the Pacific Energy Center for training purposes.  This first curve is typical for one of the two pumps that were installed for the ice storage system at the
facility.  Originally, there were two identical pumps with impeller curves that matched the one highlighted in light blue below.

Pump-curve-1

The design intent was that both pumps operate together with each pump producing 53 gpm of flow at 110 ft.w.c. for a combined performance of 106 gpm at 110 ft.w.c.  The parallel pump curve for both pumps operating together would look like the following
curve.

Pump-curve-2

When we tested the system during the first of our ongoing series of Retrocommissioning workshops, we discovered that the pumps were over-sized and that we could deliver the required flow of 106 gpm with only 58 ft.w.c. of head.  The result was that we trimmed
the impeller of one pump to provide this level of performance and left the other pump untouched. 

As a result we had two dissimilar pumps in parallel, one with the curve highlighted below in light blue and the other with the curve highlighted in orange.

Pump-curve-3

If you construct the parallel pump curve for these two pumps, you get a curve with a “funny” shape as illustrated by the dark blue curve below.

Pump-curve-4

The result, when the existing system curve is imposed on the combined pump curve, is as illustrated below.

Pump-curve-5

Bottom line is that P2 (with the smaller impeller) running alone delivers the design flow, just as we intended.  But if we start P1, it over-powers P2, delivering more flow than required all by itself while P2 is dead headed by the pressure produced by P1, providing not flow but consuming energy (and generating pump heat) any time it runs.  This would normally be an undesirable situation, but we allow it to persist at the PEC to demonstrate the phenomenon.

So, for the reason illustrated above, dissimilar pumps piped in parallel are a “red flag” for me.  Since the system I was looking at in Newport had the potential for this type of problem I wanted to see if I could detect it with my data logging effort. 

Ultimately, I decided to use two of the inputs I had available to me to monitor the current used by the HVAC hot water pumps.  My theory was that since they served a constant volume circuit, the power required and current consumed should hold fairly constant.  If it varies, then that is a clue to me that something – perhaps the interaction with the process pumps – is pushing them around on their pump curve, meriting further analysis.

I supplemented the current logging with a pump test using the existing gauges so that I could assess the operating point of the pumps at the time I installed the loggers via the pump curves.  I will discuss this in more detail in a subsequent post.

For those who are interested in the reasoning behind our decision to trim the impellers on the PEC pumps (for instance why did we do that instead of installing a variable speed drive), you will find that information and more in a new design brief titled Centrifugal Pump Application and Optimization, which can be downloaded at no cost from the Energy Design Resources web site.


David Sellers
Senior Engineer – Facility Dynamics Engineering

This entry was posted in Uncategorized. Bookmark the permalink.

3 Responses to Parallel Pumps; An Indicator of a Retrocommissioning Opportunity

  1. Andrew says:

    Thank you, I found this post very helpful, and I was busy trying to
    construct curves for pumps in parallel

  2. It’s so lucky for me to find your blog! So shocking and great! Just one suggestion: It will be better and easier to follow if your blog can offer rrs subscription service.

  3. Chris:
    There is an RSS subscription button at the top of the blog page, just above the photo of David. So subscribe away!

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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