System Diagrams: Order of Connection Matters

In my last post, I used a case study to illustrate why the “untangled” nature of a system diagram can be useful for identifying and troubleshooting field problems.  In the post before that, I pointed out that “untangling” things will only be useful if the drawing produced still retains the proper order of connection for the various elements.  The following case study illustrates why.

The example is set in a central chilled water plant serving a hospital in California and happened when I was working with Karl Stum at PECI.  In preparation for his first field trip, Karl had made a system diagram of the chilled water plant for the hospital based on the design documents, which revealed the system was a variable flow, primary secondary system arranged as illustrated below.

One of the reasons Karl had made the drawing was that the operating team had complained that Air Handling Unit (AHU) 4 and 5 would sometimes get warmer chilled water than AHU 1 – 3.  The problem was particularly mysterious because it was not always there and the magnitude of the temperature difference seemed to vary all over the place.

As you might expect if you are familiar with Murphy’s Law, AHU 4 and 5 served the critical loads in the facility (ICU and the surgical suite).  So, one of the targets of the retrocommissioning project was to resolve that issue.

Based on the system diagram, both Karl and I were inclined to think that the issue was related to calibration of the sensors, or maybe a run of uninsulated pipe in the mains to AHU 4 and 5.  We reached this conclusion because the diagram indicated that all of the water leaving the chillers was mixed in a common pipe before it got to the distribution pumps serving the AHUS.  Thus, it seemed physically impossible for temperature of the water delivered to AHU4 and 5 to be any different from the temperature of the water delivered to AHU 1-3 unless there as some source of heat down stream of the pumps and only in the piping to the AHUs in question.

That’s not to say that the temperature of the water leaving the chiller plant would always be the same as the temperature of the water leaving the chillers.  One of the characteristics of variable flow, primary secondary plants is that by design, water returning from the loads can bypass the chillers.

Understanding this is important in the context of this case study, so I’m going to digress a bit to provide a quick overview of variable flow primary secondary plant operation;  just enough (I hope) to give you the background you will need to understand the case study.  If you want to know more, I have uploaded a PowerPoint show on variable flow chilled water plant operation into the public folder of my Google Documents account that you can download are review.

I say “by design, water returning from the loads can bypass the chillers” because the fundamental concept behind a primary secondary variable flow plant is to allow prime movers like some chillers (especially older chillers which can not tolerate much flow variation in their evaporators with out danger of freezing) to operate in conjunction with variable flow loads.  The driver behind wanting the flow to vary to the loads is that it allow the pump energy associated with serving the loads to vary as the load varies.  If you don’t need much cooling, then a properly design variable flow load will use less pumping energy than it would at a time when maximum cooling was required.

As a result, a variable flow primary secondary system consists of two pumping loops, a constant flow loop serving the chillers and a variable flow loop serving the loads.  But, for water to move from one loop to the other, they have to share some pipe and the shared pipe is what allows warm return water to bypass the chillers under some operating conditions.

Specifically, when the loads are demanding more flow than the chiller pumps are circulating, then the extra flow will bypass the plant from the return to the supply side, causing the supply temperature leaving the plant to be elevated relative to the supply temperature from the chillers.  How much the system supply temperature is elevated is a function of the ratio of the bypass flow to the total flow.

  • If there is only a little bit of bypass flow, then the chiller leaving water temperature will be very close to the supply temperature and their generally will not be any operational problems associated with the bypass flow.  In fact, I have operated some older chilled water plants where the chillers were staged on based on the bypass flow rate and/or the resulting supply temperature shift.
  • If there is a significant amount of bypass flow, the chilled water supply temperature to the system can be elevated significantly relative to the supply temperature from the chillers.  A common term for this condition out in the field is “overflow” as in “the distribution pumps are overflowing the plant”.

Overflow is the Achilles heel of variable flow primary secondary plants because the elevated supply temperatures it creates can wreak havoc on the loads.  Frequently, this causes them to demand even more flow, which makes a bad situation worse.

Returning to the case study, the following diagram illustrates an overflow event in the context of the chilled water plant Karl and I were working with.

Red shading indicates water that is at the return temperature from the loads.  Blue shading indicates water that is at the supply temperature from the chiller.  Purple shading indicates water that is a mix of return and chiller supply water.  If you visualize overflow in terms of color, the more bypass flow there is, the “redder” the purple will become.  If the bypass flow is minimal, the purple will be “bluer” in hue.

Hopefully, at this point, you can see why Karl and I concluded that it was impossible for the different AHUs to get water supplied at a different temperature unless there was a problem with missing insulation on the lines to the units that got the warmer water (AHU 4 and 5).  That seemed unlikely because it would take a pretty long uninsulated piping run to raise the temperature that much and there would likely be condensation problems associated with it that would have manifested themselves as water damage, unless the lines ran exposed on the roof.

Bottom line is that Karl headed off to the site anticipating a calibration problem at the various sensors monitoring supply temperature to the loads with missing insulation being a less likely possibility.  So, he was quite mystified when, after verifying that everything was insulated, he documented different supply temperatures in the mains serving AHU 4 and 5 vs. the mains serving AHU 1 – 3 with his temperature standard.

As we were talking on the phone, it suddenly occurred to us that the one step we had not taken with our system diagram was to field verify it.  I had past experiences where a subtle difference in how things were piped created a mystery until you recognized it.  So, after we finished our conversation, Karl headed off to the mechanical room to do some field work.

He called me back a little later with the mystery solved.  Here is what his field verified piping diagram looked like.

If you compare this picture with the preceding diagram, you can see that there is a subtle but significant difference.  Specifically, in the original design, the header serving the chiller came together into a common pipe before splitting to serve pumps CHP4 and CHP5.

But the installed system connected the pumps directly to the header with CHP4 being connected between chillers 2 and 3 and CHP5 being connected between chiller 3 and the bypass pipe.  As a result, when conditions where such that there was bypass flow from the return header to the supply header, the following situation occurred.

As you can see, the “as built” piping configuration caused pump CHP5 to pull a mix of bypass water and chiller supply water while pump CHP4 saw only chiller supply water. As a result, the water leaving CHP5 was warmer than the water leaving CHP4.  And since the main serving AHU4 and 5 was connected to the pump discharge header closer to CHP5, those units received warmer water than the units that were connected closer to pump CHP4.

Note that if the bypass flow rate was low, the difference between the temperature of the water leaving CHP4 and CHP5 would not be very different.  In contrast, if the bypass flow rate was higher than the flow through pump CHP5, AHU4 and 5 might be served with water that was at the return water temperature.  With a high enough bypass flow rate, the water leaving CHP4 could even be elevated above the chiller leaving water temperature if CHP4 pulled some of the return water.

The point of the preceding paragraph is that the diagram illustrates one of many conditions that could occur given the dynamics of the system which include:

  • The number of chillers in operation
  • The number of distribution pumps in operation (CHP4 and 5)
  • The speed at which the distribution pumps operate at (CHP4 and 5 had variable speed drives)
  • The position of the control valves on the loads
  • The set points controlling the chiller leaving water temperatures, the pump speeds, and the leaving air temperatures on the loads

All of these variables made the problem very dynamic and contributed to its mysterious nature.

The bigger point in this case study is that the specific order of connection can make a huge difference in how a system works.  In this particular case, a previously undocumented field change to the way the distribution pumps connected to the header played a major role in creating a mysterious operating problem.  Piped as designed, the water would have had to mix in the common pipe between the chiller leaving water header and the distribution pump suction header.  The actual piping created independent paths between the pumps and the leaving water header serving the chiller and set up the problem.

Complicating the issue was the fact that the design and installed configuration of the pump discharge header was such that the water from the two pumps did not go through a common section of piping before branching out to the loads.  As a result, loads closer to one pump were favored by the water leaving that pump.    And in a perfect storm of Murphy’s Law corollaries, the pump that was piped to have “first shot” at any plant bypass water was also piped to favor the critical loads with its discharge water.

If you are new to the field side of things, you would think that the difference between how CHP4 and 5 were actually piped vs. how they were piped in the field would be obvious;  it jumps right out at you when you compare the two system diagrams.  But in the piping maze that occurs when you apply the physical constraints of a mechanical space to a design, things may not be as obvious as you can begin to see from this picture.

Note that this is not a picture from the plant Karl and I were working with in this case study.  That happened before digital camera’s were as common as they are now and I don’t actually have any pictures from that project.  However, the configuration of the plant was similar to the one in this picture.  Plus, this shows how “spiffy” a well maintained plant can look in addition to showing how tangled up things can start to get when you pipe them up.

In the next post or two, I will walk through the actual steps I used to untangle the piping maze at the Pacific Energy Center and develop the system diagram for the chilled glycol/ice storage system that is there.

David Sellers
Senior Engineer – Facility Dynamics Engineering

Click here for an index to previous posts

This entry was posted in System Diagrams. Bookmark the permalink.

2 Responses to System Diagrams: Order of Connection Matters

  1. Ghassan says:

    Dear Sir,
    I’m working on a hotel project consists of 15 floors. At the top, there are 6 chillers with equal capacity and 4 pumps, the chilled water distribution as per the contract is like a main header for supply and return and all the six chillers are connecting in parallel to this header. The pump set is connection to the return header. The building load also is connecting to the same header with a multiple branches going to each shafts. I am not sure about the water flow balancing in this system . can you help me in this matter please? I will send pics and drawings if you reply.
    Thanks a lot.

    • Hi there,

      Sounds like your concerns may be justified based on the description you provide; maybe a situation similar to the case study, but more complex.

      If it would be helpful, we could try to start a discussion about it, not just with me but with others who visit the site. There are a lot of good minds out there.

      I’d be happy to comment too, but as you can tell from my posting rate, I get distracted by “life” on occasion so I ‘m not too timely some times in my response, this being a case in point.

      Having said that, if you want to start a discussion, send some pictures and a diagram and we can put them up in a post for folks to comment on and then move on from there.

      In any case, thanks for visiting the blog,


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