Using Scatter Plots to Assess Building Performance–Part 7

A number of these posts have focused on a data cloud for an office/training facility located in Monterey California.  Much of the work associated with the examples I have used occurred as a part of a field class where we were using the facility for training purposes.   The class happened over a year ago now, and Brian and Jay, the two energy engineers involved on the site and who spearhead the classes there, have been hard at work implementing improvements.  

As a result, we are actually starting to see the results in the current data cloud so I thought it would be interesting to share their success with you so you can see that the indicators from early on really did lead to meaningful savings.

In the course of discussing that, I  will also talk about a few tricks I use to view before and after data.  In addition, I  will wander off for a  minute and discuss linkage system kinematics since they came into play in terms of capturing some of the economizer savings .

Viewing the Data

The data set I am using is one that has data from January of 2014 through mid September 2015 (just a few days ago).  One way to compare the two data clouds would be to filter it for the before condition, copy the graph into PowerPoint or Word or something, then filter it for the after condition, copy the graph again, and then compare the copies. 

That’s not particularly a bad thing;  after all, you probably are going to want to present your results anyway, so having them in Word or PowerPoint certainly is a useful thing. But another option that I have found to be really handy in terms of tweaking things and making quick comparisons is to set up the data set with all of the filters that you want, including a graph inserted over them,.  You can see how I go about doing that in the video included with the post titled Using Scatter Plots to Assess Building Performance–Part 6.

For me, that arrangement makes the analysis process easier and faster since I can watch the changes happen as I make them with out toggling back and forth between a chart tab and a data tab.  In any case, if I want to compare two different times in the same data set, I have found it to be very useful to make a second copy of the data set, including the chart once I have things set up the way I want.

What all of that means is that I literally have  two identical tabs in the same spreadsheet.  The only change I make  is to change the color of the fill for the markers in the chart on the second tab.  Then, I make a copy of the chart from the second tab and paste it into the first tab adjacent to the chart that is already there.  When I am done, I have a tab with two identical charts other than the color of the data set.  


But, because one of the charts is referencing the second tab (the green data set in the image above) and the other chart is referencing the first tab, I can filter them independently to view changes.

You could also accomplish this by simply adding a second series to the chart in the first tab and then formatting it differently with a different color.  If you do it this way, the clouds are sitting on top of each other, which can be useful sometimes.  If you want to separate them, you can  to reference one cloud to the the secondary axis, a technique I illustrate in the post titled Using Scatter Plots to Assess Building Performance–Part 4 where I use it to show a thermal energy cloud along side an electrical energy could to assess simultaneous heating and cooling.

Assessing the Savings

The Savings Targets

As you may recall, the team working on the facility we are talking about had identified a number of potential opportunities as outlined in the posts titled Using Scatter Plots to Assess Building Performance–Part 4 and Using Scatter Plots to Assess Building Performance–Part 6.   They included:

  1. The potential to reduce chiller run hours by improving the way the economizers worked so that during cool weather, the chiller did not have to run.
  2. The potential to improve schedules so the systems truly only ran on weekdays from 7 am to 5 pm, for  special events, and only if temperatures in the building were drifting too far out of range during an over-night or over-weekend shut down.
  3. The potential to eliminate simultaneous heating and cooling by optimizing the schedule for the large training rooms.

Capturing the Economizer Savings

Capturing the economizer savings involved making a number of improvements, including having a better outdoor air temperature reference and optimizing the hours when the economizer would be allowed  to operate; i.e. adjusting the economizer high limit setting.  Those sorts of improvements are  pretty common with economizers. 

The one that was  fairly unique for this project was related  to the arrangement of the linkage systems on the return dampers.  Because of the kinematics of the linkage system, the torque available to rotate the damper blades varied significantly over the course of the stroke.   These screen shots from a model I use in class should give you a general sense of the issue.  Here is an overview of the unit and then what you would see if you moved around to the back side and open up the side panels so we can look at the dampers, just to get you oriented.


Opened up

This first picture sets things up and gets us on the same page about what I  will be calling different things.  


Now, I have added a transparent yellow rectangle.  It lies on a plane through the crank arm that lets me draw things on it.  It’s like I magically put a piece of tracing paper there and it doesn’t really exist on the actual AHU. Note that I have turned off a bunch of layers that have things like casing elements, filters, etc. on them to let us focus on the damper actuators and linkages.

Linkage Motion

The point of this image is to show a couple of things.  One is that as the actuator shaft rotates, if you attached a pencil to the ball joint so it could draw on the yellow rectangle, it would draw a circle, as shown.  The same is true for the ball joint on the blade clip.

Another thing to notice is that as the actuator crank arm rotates clockwise, the blade clip and blade it is attached  to will rotate  counter clock-wise.  Force will be transmitted from the crank arm to the blade clip by the link, which will remain straight but which will have an angle relative to the slot in the crank arm and the slot in the blade clip that varies through-out the stroke.  That will become important and you will see why in a minute.

This next image illustrates a fundamental physical principle associated with torque.  Specifically, as the shaft rotates and the crank arm associated with it rotates, the torque can be represented as a vector that is perpendicular to a line joining the center of the shaft with the center  of the ball joint.  The purple arrow represents this vector.

Torque Vector

But in terms of force applied to the link between the crank arm and blade clip, only the component of the vector that is parallel to the link will show up as useful force to generate torque at the blade clip and the damper blade it is attached to.  Physical principles and trig allow us to resolve the torque vector into to components that will be of interest to us, specifically, one parallel to the linkage rod and one perpendicular to it, as shown below.

Torque vector components

In other words, from the standpoint of physics, its some of the toque will try to move the link and blade clip, as indicated by the blue arrow.  But the rest of the force will simply try to bend the linkage system. 

What is interesting about this, and the root to the unique economizer problem that this particular system had, is that because the angle between the link and a line from the center of the damper shaft  to the center of the ball joint varies as the crank arm rotates, the force available to move the damper blade also varies.  In the arrangement illustrated above, the maximum force available to rotate the damper blade occurs at mid stroke because at that point, the entire torque vector is  perpendicular to the line connecting the center of the actuator shaft with the center of the ball joint.

Torque vector components - end

That is not always the case.  In fact, linkage systems can be set up (and are set up) so there is little if any torque available to the damper system at some points of the stroke.  That means that if there is any load  at all on the damper linkage system due to the air flow and/or the pressure drop across the blades, then the actuator may not be able to move the damper and it will stall out. 

The human analogy is when we put a small pipe wrench on a pipe that has become severely corroded and try to loosen it up.   We an apply a lot of force and hour heart and body is doing a lot of work.  But the force we are applying through the wrench to the pipe does not do any useful work in the context of causing the pipe to rotate.  So, we either need to find a person who works out more than us or a longer pipe wrench to give us more mechanical advantage if we want to break the pipe loose.

Why This Mattered

For the systems in the facility we are discussing, the linkage arrangement was such that there was very little toque available to move the return damper at one portion of its stroke.  As a result, as things changed and the flows and pressure drops through the dampers changed, the systems would reach a point where the force on the blades over-powered the force available from the actuators and the actuators would stall out and not move the return dampers.   Kind of like us straining against a corroded pipe fitting with a pipe wrench that was to small for the job.

This impacted the economizer’s ability to provide cooling in addition to creating some fairly significant (and dangerous) negative pressures in the mixed air plenum (negative 4 to 5 inches w.c.).   In fact, that is how we noticed the problem initially. 

Specifically while working in the mechanical  room, we noticed that the systems seemed to generate a lot of air noise like the sounds you get when air  is leaking through cracks.  Turns out that was happening because the dampers had been stalled out and negative plenum pressures were sucking as much air as they could into the system  through all of the door seals, compartment flanges, etc.

Steve, the mechanic serving the facility was able to quickly solve the problem by reconfiguring the actuator system once we realized it was there.  It’s a perfect problem for a mechanic when you think about it;  its all about levers and angle and linkages, the type of thing mechanics love to figure out and fix.

The Economizer Savings

Most of the changes to that were  targeted were made during April and May of this year (although the linkage improvement to the economizer happened during the class last July).  So, to get a feel for how successful the improvements were, I filtered the data to look at consumption for June, July, and August of 2014 in one data set and to look at those same months but in 2015 for the second copy of the data set.   Here is what that looks like.


When I looked at the results, my first reaction was to the visual change; i.e. the shape of the clouds was clearly different.  But when I looked at the total kWh for the two periods, I was a bit disappointed because while there were some savings, they were fairly modest (75,908 kwh for 2014 and 72,600 kWh for about the same period in 2015).  But then I realized a couple of things.

One was that the a lot of the economizer savings would occur for hours when it was cool enough for outdoor air to meet the system leaving air temperature set points.   For these systems, that means outdoor air temperatures at or below the mid 50’s°F.   For the time period since the improvements were made, there have not been many hours where those benefits would accrue.   So, we will have to wait and see what happens when cooler weather sets in.

The other thing I noticed was that the weather for this year relative to last year has actually been warmer over the course of the three months we  are looking at.  So it is likely that the chiller had to run more often than it did for the same interval in 2014.  So the fact that there still were savings probably meant we had made some progress, probably related to some of our other targets.

Schedule Improvement Savings

A second area where we thought we would see electrical savings was related to scheduling, specifically by eliminating what seemed  to be random system operation during hours when there really should not have been anything going on in the facility.  So, I filtered the data to eliminate the base load and the normal operating hours, just like I did to assess the problem initially.  But in the “after” period, I had a much different result, as illustrated below.


Good news;  the cloud associated with operation during hours when we really did not expect anything to be operating became much smaller.  And the hours that did show up are at a lower level of demand, so they may represent variations in the true base load rather than system starts. 

I could probably hone in on that some more by filtering the base load at a  bit higher value combined with some trend data for system operation to help me sort it out.  But in the big picture, we are clearly making progress.

Having said that, if you look at the savings, it is not as significant as we anticipated.   This puzzled  me a bit so I took a look at the data as a time series vs. a scatter plot and here is what I saw.


The historical data says that there was something going on that tended to shift the consumption pattern up during the cooler months, starting in mid September and possibly persisting through mid June, based on the 2014 data.  It shifts down again during the weeks around the winter holidays, probably because the building is scheduled off at that point.

I can’t really prove it with this data set, but one way to explain such a pattern would be dysfunctional schedules and night set-back features, which might run systems more at night than necessary once the set back cycle is triggered.  We (as a company) are seeing this fairly frequently and we think it is related to a subtlety of how a lot of DDC systems  do night setback.  It is a long discussion and I plan to do a separate blog  post  to explain it  once I have a bit  more data to work with.

My point in bringing it up here  is that it is a likely culprit for driving consumption up during cooler months and I suspect it may have been going on here.  Brian initiated the most of the changes in April and May of 2015, as I mentioned previously, and clearly you can see the impact of that in the time series data.  

But in the “after change” period we don’t have data beyond mid September; i.e. no data for cooler  weather.  So this is another one where time will tell if the savings we anticipated are fully realized.  But the data pattern already is telling us that one of the problems we targeted has been addressed.  And it also likely reflects the electric side of the simultaneous heating and cooling savings, based on what I discuss next.

Simultaneous Heating and Cooling Savings

As you will recall we were hoping for significant thermal savings via improvements to the way that the training rooms in the facility were scheduled.  It turned out that capturing the savings was a fairly straight forward implementation.   The thermostats already had a manual over-ride request button; i.e. a button you could push to generate a signal that the DDC system could use to trigger an occupied cycle.   The problem was that nobody in the facility knew about it let alone how to use it.

In addition, the system serving the meeting room was scheduled to run round the clock so even if someone did know what the button did, for the meeting room, they would have only needed  to use if if the meeting room was in use after hours.

To capture the savings for this target, Brian did two things.  One was to program the meeting room system so that it only ran if someone pushed the override request  button on one of the thermostats associated with the four zones in the area.  Otherwise, the system does not run unless it needs to come on line during extremely hot or cold weather to keep the space temperatures from drifting too far  out of spec (i.e. night set-back and set-up).

The other thing he did was simply train the staff about how the thermostats work, including making little signs that mount around the thermostats and describe how to use them. Her e is a picture of one of his signs for a conventional zone.  Its very similar to what he did for the meeting rooms, but the meeting room signs include a few meeting room specific details, like where the thermostats are located.

Thermostat Label

I suspect I can procure a picture of the actual meeting room signs and put it up here.  But I think it’s a really great example of how to help the folks using the facility understand how they can make a difference in its energy use patterns.   Many of the people I run into are just as concerned about energy use and the environment as I am.  So helping them understand how they can do their part, like Brian and Jay did with their signs, will typically pay off.

What I do have a picture of is the change in the thermal cloud, which is pretty significant and what we expected.


Clearly, the improvements to how the meeting rooms are scheduled have generated meaningful savings on the thermal side.  And so far, we are also seeing some electrical savings as discussed above.  But I anticipate that the bulk of the electrical savings are yet to be seen for reasons as indicated.  So I will be keeping an eye on things and try to report back again on this facility next spring.

Meanwhile, have fun looking for shapes in the clouds.


David Sellers
Senior Engineer – Facility Dynamics Engineering

Click here for a recent index to previous posts

This entry was posted in Excel Techniques, Retrocommissioning Findings. 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