This post picks up where I left off in the string of posts I started about a make up air handling system that was simultaneously heating and cooling when it didn’t need to. And, since I finally figured out how to hyperlink to headers inside a WordPress post  thanks to a well written blog post by Wendy Cholbi, I will now be able to include a list of topics at the beginning of my posts so you can jump directly to the content of interest if you don’t want to read the entire thing.
For this post, the topics I will discuss are as follows.
- Why I Decided to Log Data
- Taking a Data Snapshot
- Formal Monitoring Plans
- Informal Monitoring Plans
- Changes in Enthalpy = Changes in Energy
- Changes in Temperature = Changes in Energy if the Coil is Dry
- Short Term Temperature Swings Provide Long Term Insights
- Humidity Eliminates the Simplifying Assumption
- Logging Air Flow
- Logging Water Flow
- Logging Air Side Pressure Drop as a Proxy for Flow
- Locating the Temperature Sensors
- Stratified Temperatures = More Loggers
- A Chance to Test your Skills
Why I Decided to Log Data
As you will recall, a number of clues had added up causing me to suspect some focused attention on this system would yield benefits in terms of energy savings, improved performance and reduced operating and maintenance costs.
- The system had the potential to simultaneously heat and cool by the nature of its design.
- The system handled large quantities of outdoor air by design.
- The system needed to run round the clock to accomplish its mission.
- There were a number of similar systems in the facility and the facility showed a high baseline utility consumption pattern, something that could be contributed to by dysfunction in the makeup air systems.
- The system used pneumatic controls which were in need of attention.
- The facility operators felt there were problems with the system based on their operating experience.
- There were a number of obvious things that were wrong with the system, including that it was actually simultaneously heating and cooling unnecessarily when the Chuck, the engineering supervisor and I walked up to it.
Because of all of the clues, I decided to deployed a data logger on the unit so I could capture a day or two of data while I was on site to help me with my initial assessment. I carry several data loggers with me all of the time for situations like this.
Taking a Data Snapshot
While a true analysis of the system performance would require collecting several weeks of data at a minimum, picking up a couple of days or even hours of data while you explore other areas of the site can:
- Provide additional insight into a problem,
- Confirm your suspicions, and
- Provide tangible evidence of the need to move forward with improvements.
Formal Monitoring Plans
For most projects, logger deployments will occur after the initial scoping/assessment phase and you will have spent some time thinking through what you need to know. In fact, you probably will probably develop a monitoring plan of some sort. That is what ultimately happened at this facility as the students in the class we were teaching there moved forward with their projects. I have placed one of their monitoring plans on my Google Drive if you would like to take a look at it.
The bottom line is that a formal monitoring plan simply documents your intentions in terms of data logger requirements, sensor requirements, sensor locations, etc. To develop it, you address the same items I am about to discuss in the context of the informal monitoring plan that I developed for this project.
Informal Monitoring Plans
In a field situation like the one I am describing, even though you don’t have a formal monitoring plan, you still need to come up with an informal plan of some sort. For me, that means asking myself what I really need to know to get some perspective on the issue. This is time well spent since it will ultimately inform the development of the formal monitoring plan later on in the process.
In this case, I suspected the system was doing unnecessary simultaneous heating and cooling.  So,to gain some insight into that, I asked myself how I would go about calculating the energy waste if it actually existed. The answer to that question was that I would need to know the change in energy as the air moved through the air handling unit.
So, it’s that simple; I just need to log or somehow measure the parameters that would let me assess the energy changes in the air stream as it moves through the air handling unit.
Logging the Change in Energy Across the Air Handling Unit Coils; Changes in Enthalpy = Changes in Energy
As you may or may not know, the energy changes in the air stream come back to the change in enthalpy across the heat transfer elements in the Air Handling Unit (AHU) and there is a fundamental HVAC equation that will tell you the answer, assuming you can provide the inputs.
The image above is a screen capture from a set of slides I use occasionally called Useful Equations for Basic HVAC Calculations. It’s a set of slides that has HVAC equations that are commonly used out in the field and in design to quantify basic parameters like flow, temperature change, moisture change, energy change, pump and fan energy and power, etc. I have a copy of the slides posted in the public space of my Google Drive if you are interested. You can also find this equation and others like it in Arthur Bell’s HVAC Equations and Rules of Thumb book., which you can now get for your Kindle.
Returning to the equation above, until you become accustomed to the term, enthalpy can be a little scary. For what its worth, in college, I almost flunked thermo because it took me a while to get the idea behind entropy and especially, enthalpy. I want to emphasize that this was not the fault of the instructor, Mr. Flanagan, or “easy Ed” as he was affectionately referred to.
This was not because he was “easy” in terms of setting expectations for you. It was because he had this very quiet, soothing voices, so his lectures were kind of like listing to the “easy listening” channel in terms of how they sounded. But the content was really solid physics, in this particular case, thermodynamics.
To make sure you “got it”, he had an open door policy that I took advantage of frequently. During these sessions, Mr. Flanagan would patiently go over the nuances of lecture, many, many times, in many, many ways if need be. And eventually, I got it.
So, fear not; basically, enthalpy is a term that references both:
- The energy that is associated with sensible heat, which is energy we, as humans detect as a warmer or colder temperature, and
- The energy that is associated with latent heat, which is energy we, as humans, detect as a humidity.
So, if you know the enthalpy change across a coil and the flow rate, then you can calculate the change in energy across the coil.
As you might suspect, logging the enthalpy change across a coil involves logging multiple parameters; specifically some indication of temperature and some indication of moisture.
Moisture, as you might imagine, is much harder to measure than temperature. I could rattle on about that, but the Iowa energy center did a lot of research to document the challenges and the best solutions. So, you can learn more by linking to their reports on humidity sensors, blind tested “out of the box” and also over time.
The bottom line is that humidity is much harder to accurately measure than temperature, a problem you can eliminate by making a simplifying assumption.
Changes in Temperature = Changes in Energy if the Coil is Dry
Simplifying assumptions are somewhat common in applied science, at least in my experience, and especially out in the field where I spend a lot of my time. But, its important that the assumption be valid. Assuming that gravity does not apply for a process that occurs at or near the surface of the planet is not a valid simplifying assumption. But assuming that the process in question might only involve a temperature change and not a moisture change might be.
In this particular case, the simplifying assumption is that if the coil happens to be operating under conditions where there is not a humidity change, then the equation for energy change across it simplifies to some extent. The simplification is that now, you can determine the energy change by simply monitoring the temperature change, not the enthalpy change.
Since I was on my site visit and would initially be logging data in December and January, it would be unheard of for dew point temperatures to be anywhere near the temperature of the chilled water supplied by the central plant in the facility, as can be seen below.
If the outdoor air dew points are below the coil surface temperature, which, of course, is set by the temperature of the chilled water, then there will be no condensation on the coil and all of the energy transfer will show up as a sensible temperature change in the air stream. Thus, I could calculate the energy transfer by simply logging temperatures and applying the sensible heating and cooling load equation above.
Short Term Temperature Swings Provide Long Term Insights
If you study the temperature data in the graph above, you will notice that even though the data is from the dead of winter, there are some pretty big temperature swings and some fairly warm days encountered during the interval. I actually love getting to log data through an event like that.
For one thing, when Mother Nature does that, the system is being exercised in a very realistic natural response manner. Its much better than any forced response test that I could come up with. And if there are issues with the control processes or system or control hardware, they are likely going to show up. during an event like this. These slides from a recent webinar I did illustrate the point in the form of a real event happened in the Midwest one day while I was working at client site and got to watch their surgery system deal with it all.
The bottom line is that over the course of one 24 hour period, a weather system moved through that put the unit through every single possible operating mode. The particular system I was watching handled it pretty well, but a lot of thought had been put into just such an eventuality in the design of the control system and the selection of the hardware.
And the operating team was top notch and kept things well maintained and knew what to watch during an event like this. I have been around a lot of other systems that would have never made it through such a transition with out a bunch of problems cropping up, including multiple freezstat trips, frozen coils, and loss of service to a critical area.
I bring this up here because if you get to log data through big climate changes, then you perhaps are going to learn even more about the system that you set out to learn. In our case, we set out to see if it was simultaneously heating and cooling, but if there happened to be some big temperature swings, which is the norm for the area in at that time of year, we will also learn about how robust the control system and hardware are.
The other thing that the large temperature swings do is characterize the performance of the system over a larger temperature span. That means that if we end up wanting to do some sort of regression analysis to predict energy consumption as a function of outdoor air temperature, we will have a better data set to build our model from. I will illustrate that a bit more in one of the subsequent posts.
Humidity Eliminates the Simplifying Assumption
In Denver, our simplifying assumption would not work so well in the summer, as can be seen from this graph of the outdoor conditions for the area from mid June through mid August.
That means that if I needed to extend my study into the summer months to more accurately document what was going on, I would need to start logging humidity in addition to temperature at some point if I wanted to really calculate the energy change across the coil.
Or maybe not;
The first law of thermodynamics states that the total energy of an isolated system will remain constant and while the energy can be converted from one form to another, it can neither be created or destroyed. Stated more simply, as Dr. Al Black, one of my mentors would often say:
The goes intas gotta equal the goes outas.
Meaning that one trick you can use to pick up an energy change that occurs across a cooling coil where both sensible and latent cooling is happening is to measure the temperature change on the water side and then use the water slide load equation, as illustrated above.
Of course, to apply the water side equation, you have to be able to measure or reasonably estimate the water flow. The same is true for the air side calculation, all of the air side equations have air flow as a parameter.
Logging Air Flow
Logging flow is much more challenging that logging temperature or status. But its not impossible. One technique would be to use an arrangement similar to the one I discussed in my post about logging filter performance. That involved using a 4-20 ma velocity sensor, a power supply panel, and an input cable for the Hobo logger that used a resistor to convert the current to a voltage that was acceptable to the logger. The links take you to the details of all of that in previous posts if you want to know more.
In addition, Onset (the manufacturer of the Hobo logger I was using) offers a growing array of products that would let you pick up flow with an off-the-shelf sensing system you could purchase directly from them (note that you typically need a cable or two and a power supply in addition to the sensor). Bottom line is that logging flow is possible and becoming easier and easier as technology advances. But technology comes at a price and, if you compare the cost of a temperature probe and something to measure flow, you will find a flow point is much more expensive. That doesn’t mean you shouldn’t do it. But it is something to think about as you develop your project budgets.
In the case under discussion, I was able to solve the problem by making another simplifying assumption, made possible by the nature of the system. Specifically, the system was a constant volume system, meaning the design intent was that it would generally deliver the same flow rate no matter what was going on.
The reality is that most constant volume systems are only quasi-constant volume. For example, as the filters load the fan will be pushed up its curve unless there is some sort of variable speed drive and related control strategy in place to hold things steady. Another thing that can shift he flow around is the movement of dampers in the system, something I discussed and documented in a previous post on testing an economizer for temperature and velocity stratification.
So assuming a constant volume system is constant volume might be a reasonable assumption but not a perfect assumption. In our case, it was a pretty good one because:
- The filters were relatively low efficiency and thus not as significant of a pressure drop in the system as they might have been if they were high efficiency filters.
- We would only be logging data, at least initially, for a week or two, so any change in filter pressure drop would be modest. 
- The system was a 100% outdoor air system, meaning it didn’t have any economizer dampers to move around and change things. In fact, it didn’t even have the intake damper that was shown on the project documents, which could explain some of the frozen coils.
So, bottom line, I used my trusty four-in-one instrument …
… to take a set of velocity readings across the intake louver and plug them into my spreadsheet tool …
… to come up with a general idea of the flow rate.
Lacking that, it would also have been reasonable to assume the design flow rate. But a measured value (assuming you use proper technique and get good data) will usually be better than taking information from the documents. In this case, the actual flow, based on my readings (and later confirmed by a pitot tube traverse) as 2,141 cfm. In contrast, the design documents indicated a flow rate of 3,152 cfm.
That’s a fairly significant difference and certainly would have impacted our savings projection by overstating what was actually going on. The difference also helped to explain the negative pressure problem they were experiencing in some areas served by the system.
Logging Water Flow
I mentioned that it would also be possible to measure water flow and temperature change as a way to understand the load on a coil that was condensing with out the need to log temperature and humidity on both sides of the coil. Since I didn’t do that, I won’t go into a lot of detail about that approach right now other than to say that measuring water flow can be just as challenging as measuring air flow.
And, even though the systems serving the coils in the air handling unit were constant volume, the flow through the coils will vary as the valves move to match the delivered capacity to the requirements of the load. So, making an assumption of constant volume on the water side would simply not work and you would have to log data somehow.
One trick I have used in the past to do this is to log pressure drop across a balancing valve and then use the flow coefficient of the balancing valve to calculate the flow indicated by the pressure drop. Here is a picture of us using that approach on the heating hot water system at the Pacific Energy Center a while back.
The hoses go to 4-20 milliamp pressure transmitters …
… and become inputs to the logger in a manner similar to what I described above for the air flow sensors. So, not as cheap as logging temperature, but doable.
Logging Air Side Pressure Drop as a Proxy for Flow
Incidentally, I have also used a variation of this technique to log flow on air handling systems. Specifically, I have logged differential pressure across a fixed resistance of some sort in the system, like a coil. Then, I use the logged data and the square law to calculate flow. To do that, you need to take one reference flow reading at a known differential pressure so you have something to base the calculation on.
And, if you use a coil, a heating coil is better than a cooling coil because the pressure drop on a cooling coils changes when it get wet, all other things being equal. Filters are not a good choice, at least for a long term deployment, because their pressure drop will change over time as they load up.
Locating the Temperature Sensors
A system diagram is a great tool to help you decide where to locate your temperature sensors or any sensors for that matter. The graphics below are the system diagram I developed for the make-up air system on the project, both an over-all view of the system … and a close-up of the air handling unit we are focused on in this discussion.
…and a close-up of the air handling unit we are focused on in this discussion.
You can find a .pdf copy of this diagram in the public space on my Google Drive if you would like to have a closer look.
Having said that, at the time I was contemplating the logger deployment, I had not developed the system diagram. But truth be told, it doesn’t take long to sketch something up for the air handling unit, and that sketch can become the starting point for the full blown drawing, along with the existing project documentation and some field work.
In any case, since I was trying to understand how much simultaneous heating and cooling was going on, I needed to have temperature sensors deployed across each cooling coil, not just across the collection of coils. In other words, if what I thought was going on was actually happening, the system was taking outdoor air, over heating it, and then that overheated air was going through the cooling coil, which was taking the temperature back down again.
So bottom line, I would need the temperature coming into the unit, the temperature after the preheat coil, and then the temperature after the cooling coil. I could have minimized the number of logger points I needed by finding a local weather station and pulling the outdoor conditions for the area from that location, which would mean that I would only need to log two temperatures in the unit.
But, climates can vary in short distances, so picking up the data locally with my own logger would paint a better picture if I could do it. Plus, I could calibrate all of the sensors relative to each other before deploying the logger to eliminate any false temperature differences, just like I did in my tea kettle experiment.
At the time, I had a 4 channel Onset U12-013 logger at my disposal, which had an internal temperature and relative humidity sensor along with two external inputs. Since I also had the temperature sensors with me that could be used with the external inputs, I was pretty well set up to pick up the information I needed with out having to use a local weather station. So, as a starting point, I decided I would physically locate the logger itself in the entering air stream so that it’s internal sensors could pick up the local climate conditions.
There were no access doors into the intake plenum, but Chuck and I were able to pull out a couple of filters until we had one from about the middle of the unit and then snuggle the logger into the pleats. Then, we slid the filter back into the rack. Not pretty but it worked.
You will recall that one of the things that had attracted my attention about this unit before I even was on the site was its configuration, meaning that the coils were literally installed the way I showed them on my system diagram, which is a less than ideal arrangement in terms of providing a functional preheat cycle. Specifically, with the coils literally installed face to face, there is little room to install a freezestat and controlling sensor downstream of the preheat coil and ahead of the cooling coil. 
And, upon encountering the unit in the field, I discovered that my concerns were warranted; the coils were indeed installed face to face and there was evidence of freeze-ups, as I discussed in a previous post.
So, our challenge was to see if we could get a probe in between the two coils. By taking the side panels off of the unit, we found that we could feed a probe through a small hole that was in the coil frame, dangling it in the air stream between the two coils. We used a similar approach for the probe that monitored temperature leaving the chilled water coil, which is pictured below.
We did not dangle the probes too much because we didn’t want them to move in the airflow and touch either coil since that would throw the reading off a bit since the probe would be sensing the actual surface temperature of the coil vs. the temperature of the air stream. Once we thought we had things about right, we put some tape on the cables to keep them from shifting and put the panels back on the unit and tightened them up.
That last step is important, even thought it takes some time and makes it harder to get the logger back out. But if you didn’t put the panels back on, or put them on but didn’t tighten them down, then air from the equipment room could leak into the unit, which would mean your test data would not reflect what was really going on if there were no air leaks.
Stratified Temperatures = More Loggers
In this particular system, there was very little to stratify the air as it went through the unit. That meant that I could get a pretty good feel for the temperature everywhere by logging a single point. This particular unit was too small to get inside of and make a good picture of that, but here is an example from a larger make-up air system I was working with a couple of weeks ago.
There is a bit of variation, with the warm spot created by some leakage from the door plenum door into the equipment room, which was on the left side of the unit relative to the temperature and velocity matrix. The intake louvers were about 10 feet behind me and were large blades that tended to direct the air up, which is why I believe the velocities tended to be higher on top than on the bottom.
My little spreadsheet tool calculates the mass-weighted average of the temperatures I enter in each cell, and in this case, it was 61.2°F. So knowing that, and knowing that there was nothing upstream to change the pattern, I would have felt comfortable using a single sensor located in the middle of the array, where the temperature is 60.9. Sure more sensors would be better, but if I didn’t have them, I would get reasonable results using the location I selected based on the results of my little test.
Contrast the pattern above with the patterns we documented in an economizer down in the Bay area last January (the link takes you to the post, which has links to bigger versions of the pictures).
The three different patterns were generated by going from 8% outdoor air to about 31% outdoor air to about 83% outdoor air (left to right). And, even with out reading the numbers, you can tell from the colors that the pattern shifts radically, totally flipping by the time we get to 100% outdoor air. In a situation like this, certainly you would want multiple sensors logging concurrently across the plenum, probably at least 4 at a minimum and more would be better.
For the system we are talking about, I knew I didn’t have anything that would produce stratification. And, since the access was too tight for me to crawl in and check the pattern to inform my decision about where to put the sensor for the incoming air stream, I used engineering judgment and selected a location towards the center of the unit.
A Chance to Test your Skills
So that is the background information behind the logger deployment I did as a next step in my analysis of the errant make-up air handling system. Part of the reason I am developing this string of posts is to support some of the classes I teach. One of the things we have been doing in some of them is handing out a “thing” we call a knowledge assessment. Its basically an open book (most of the time) do it at home (most of the time) quiz.
For may of the classes I do, there is a lot of self study involved. So the idea of the knowledge assessment was two-fold:
- For the students, the assessments highlighted the areas we felt were important and gave them an idea of what they should carry away with them.
- For us (the instructors) the feedback from the assessments told us how well we were (or were not) doing in terms of communicating the ideas.
Recently, a number of students commented that they learned a lot from the knowledge assessments because it made them apply the principles. And then, the detailed answer keys we handed out provided a lot of insight into the solution to the problems.
So, for some of the assessments, this string of blog posts, and other like them will become the answer key. That means you can “play along” so to speak if you want to.
Towards that end, I have posted the system diagram and the logger data for the system on my Google Drive. I have included both the original Hobo data file and an exported .txt version of the data. TXT stands for TeXT, which is a very basic file type with little if any formatting. Excel can open that type of file and also delimited files like .CSV (Comma Separated Value) files, which is something I discuss in the linked blog post.
The advantage of a .txt file in terms of having Excel open it is that when Excel sees the .txt file extension, on most systems, it will automatically launch its import wizard, which allows you to specify the data delimiter, all of which I discuss in the linked post. Meanwhile, if you have Hoboware or the Universal Translator (not the Star Trek kind) you can work directly with the Onset files if you want.
Bottom line, the idea is for you to take the data and work with it and as you do that, consider the following questions:
- Do you see any issues? Or was my suspicion of simultaneous heating and cooling not correct?
- If you think you see an issue, calculate what you think the savings would be for the logged interval if you were to resolve it. Assume that the chilled water plant net efficiency is 1 kW per ton, that the boiler efficiency is 75% and that the boiler is gas fired. Also assume that electricity costs $0.0755 per kWh and gas costs $0.36 per therm, data you would have gleaned from the utility bills used for the average daily energy consumption analysis.
- To calculate the savings, can you use the equations as presented above? Or do you have to make an adjustment to them before using them? (Clue; think about Denver’s nick-name).
- How do the utility rates on this site compare to the rates on project sites you work at? If there are differences, how might they impact the conclusions you reach and the recommendations you make?
- Assuming you have identified an issue, what are the steps it would take to address it?
- How do you think you could accomplish the steps you identified above that are required to address the issue?
Moving forward, I will publish posts that answer those questions and (hopefully) illustrate the techniques behind the answers in doing so. Meanwhile, the hope is that this format will give you an opportunity to try your hand at the process before you jump forward to the answers. Who know, you may come up with some better ideas than the ones we came up with in the course of our project at the site.
Senior Engineer – Facility Dynamics Engineering
- One additional point to what Wendy wrote that applies if you are working in Windows Live Writer to develop your blog off-line. That is, the links won’t work until you upload them to your WordPress blog, at least that was the case for me. That threw me off for a bit since I was testing them inside Live Writer and thinking something was wrong when they didn’t work. Then the light bulb came on and it occurred to me that the post might actually need to be up on the page for the links to work. Sure enough, as soon as I sent it to the blog as a draft and looked at it in the Preview window, everything worked. (Back to content)
- There are actually legitimate reasons to do simultaneous heating and cooling in an HVAC system and a reheat process is a common example. I will make the details of why you might use what seems to be an energy wasting process the topic of a separate blog post. In this case, the simultaneous heating and cooling that was going on was a preheat process followed by a mechanical/chilled water cooling process. Not much sense in that. (Back to content)
- The clue about this is that the coil is dry, which is virtually always the case in a heating process and can be the case in cooling process, especially in places like Portland, Oregon or San Francisco, California or Denver, Colorado, especially in the winter months. (Back to content)
- That, of course, is barring some unforeseen natural disaster like Mount St. Helens erupting and spewing all kinds of ash into the atmosphere; at one point during my KSA days, I had an interesting discussion with some of the guys at Intel in Hillsboro who had been running one of Intel’s fabs back when that actually happened and had to deal with it. (Back to content)
- Recall that the preheat coil is supposed to take sub-freezing air and bring it up to a temperature above freezing to protect the rest of the system with out freezing itself in the process. That means it has to be piped and controlled in a way that will prevent it from freezing and one of the requirements in that regard is to have a temperature sensor immediately downstream of the coil to provide an input to the control process. There are a couple of ways to get around that if you have to, like using a feed-forward control process that sets the valve position based on incoming temperature and other parameters. But you still will need a sensor downstream of the coil if, for no other reason than to catch an energy wasting malfunction like the one we are discussing here.(Back to content)