Assessing Steam Consumption with an Alarm Clock

Greetings after another long absence from posting. Life got busy on me again, but I just got past the critical end of the year deadlines that were driving me. So, I figured I’d share something I learned a while back and used recently on a project.

Somewhere in 1979 or 1980, I was doing the field work for what would become my first energy audit, which was focused on Memorial Hospital of Carbondale, Illinois. One of the things we were struggling to understand was what the boiler system load profile looked like. We had gas bills, but they reflected net gas consumption on a monthly basis, which gave us a gross picture of what was going on but not a picture of the day to day load shape.  The metered gas consumption also included the inefficiencies in the boiler firing process, which turned out to be significant,  as well as gas used by the kitchen and an incinerator.

One day, while I was down at the hospital trying to understand the boiler load, Chuck McClure (founder of McClure Engineering, and the guy who ultimately took a chance on hiring an airplane mechanic/aircraft maintenance engineer  – me – after some persistent prodding by Al Black) called and suggested that I go down to the local Walgreens and buy some 120 volt alarm clocks so we could get a handle on the boiler load profile.

Specifically, he told me to buy one clock for each condensate return pump and feed water pump and that once I had them, I should wire them across the 120 vac starter coils in the various control pump control circuits, set them for 12 o’clock and then write down what they said to the second every hour or so.

He also told me to measure the condensate pump and feed water tanks paying particular attention to the level at which the float switches started and stopped the pumps and to note how long they ran when they cycled. For the feed water pumps in addition to the information he requested for the condensate return pumps, he also asked me to do a pump test to figure out where the feed water pumps ran on their pump curves.

At the time, this sounded a bit crazy; after all how could an ordinary alarm clock reveal steam consumption. But in hindsight, I realize it’s pure genius and an example of how simple information and techniques can reveal complex building operating patterns.

Here is where Chuck was headed with his request, much as he explained it to me when I got back to the office and asked about it.

The hospital had started out as a small local community clinic in the 1950’s and expanded over the years by adding wings.  Then in the mid 1970’s a new central boiler plant was installed as part of a major expansion, eliminating the boilers that had previously existed in each wing.  As a result, each wing had its own condensate pump and Chuck’s technique allowed us to use the condensate pumps to assess the load profile in the wing.

For the condensate pumps, the operating cycle was triggered by a high level float switch and terminated by a low level float switch. So, each time a pump operated, a known volume of condensate was discharged, specifically the length times the width of the receiver times the difference in float switch settings.

Since the pumps discharged to a vented return, the pressure they worked against was fairly constant, which meant their performance was fairly constant and repeatable.  In other words, the length of their “on” cycle was very consistent and repeatable.

The bottom line was that:

• If you knew how many minutes the condensate pumps ran (information provided by the alarm clock), and
• If you knew the typical cycle length (information provided by my field observation), and
• If you knew the volume of condensate discharged per cycle (information provided by my field observations and measurements),

then you knew in general terms, how much steam the loads they served had used, assuming minimal leakage in the steam system and that the loads all returned condensate.

Thus the condensate load seen by the condensate pumps was a pretty good replica for the load profile associated with the loads they served.

Condensate from the various wings was returned to the central boiler plant and collected in a deaerator;  a device that was designed to drive dissolved oxygen and other gasses from the returned condensate as a corrosion control measure to protect the boiler and steam system as well as preheat the boiler feed water.  Part of the package included pumps that took water from the dearator and pumped it into the boilers; i.e. the feed water pumps.

For the feed water pumps, the start and stop cycles were triggered by float controls on the boilers on the two boilers.  The levels that triggered the pump operation were fairly repetitive, just like the levels in the condensate receivers. But there were multiple boilers, which “muddied the waters” compared to the condensate pumps;  an alarm clock monitoring feed water pump operation would not directly know which boiler had requested it.

Since the boilers fired to maintain a fairly constant pressure, the feed water pumps, when they ran, saw a fairly stable operating condition. Thus the point were they operated on their pump curve was fairly consistent.  As a result:

• If you knew how long the pumps ran (information provided by the alarm clock) and
• If you knew where they ran on their curve (information provided by the pump test)

then you could figure out how much feed water they pumped into the boiler.

Water pumped into the boiler quickly became steam leaving the boiler. Thus the feed water load profile was a pretty good replica for the boiler load profile.

Some of you are probably thinking something along the lines of:

If you knew the condensate flow coming back from the loads, why bother with assessing the feed water flow into the boilers;  shouldn’t they be the same?

The answer to that is, like with most HVAC stuff, “it depends”.

It’s not unusual for steam to be used directly in a process and not returned to the plant.  In the hospital I was working on, many of the air handling systems used direct steam injection for humidification, which represented a significant load during cold weather with the high outdoor air volumes required by code for the facility.

In addition, the kitchen had some steam kettles that dumped their condensate to drain.  Finally, there were some leaks in the piping system. All of these items represented steam that was generated but not returned to the boiler with the humidification load in the winter being the biggest component.

These loads have some significant implications beyond the energy required to serve them.  For one thing, the steam loss they represent equates to a need to bring raw water into the system.  Raw water requires more chemicals to treat it and must be preheated from the city water supply temperature rather than from the condensate return temperature and thus represents an additional energy burden.

By measuring both the return condensate back to the plant as well as the steam leaving the plant, we were able to identify the magnitude of the direct use loads relative to the total load.  In turn, this allowed us to:

• Assess both the water treatment and preheat requirements/costs associated with serving them
• Assess the humidification load by comparing the data for warm weather (above about 55°F when no humidification was required in the local environment) with data for a cold, near design day.
• Assess the leakage in the system by comparing warm weather data for hours when the kitchen was in operation with hours when the kitchen was not in operation.

The technique Chuck shared with me has served me well over the years.  One of its beauties, at least to me, is that its so fundamental in nature; as Al Black would say “the goesinta’s gotta equal the goes outta’s”.

Lately, I have used data loggers with CTs measuring pump amps instead of alarm clocks but the concept is still the same.  Here is a Hobo U-12 logger with and external CT deployed in a motor control center to monitor current and thus operation for a condensate return pump.

The CT is the black square around the red wire towards the top of the picture and the logger is the light tan square velcro-wrapped to the Motor Control Center (MCC) structure towards the bottom.  The logger is actually deployed in the wire gutter of the MCC, which is, in my opinion, a safer deployment than one that involves opening the starter compartment.  This is 480 volt gear and I’m comfortably nervous around it, which I consider to be a good thing vs. not feeling nervous.  By working in the wiring gutter, I feel I am less exposed to the risk of coming in contact with a live part as compared to working in a starter compartment.

That’s not to say you can afford to be casual working in the wire gutter.  As you can see, there are still exposed terminals, some of which are at 480 volts.  Thus, you need to take appropriate precautions and, if you are not qualified or comfortable doing the work, have someone who is do it for you.  This is one area where taking a few extra minutes or spending a few extra dollars can be a life saver by avoiding exposure to an arc flash incident.

Getting from condensate pump amps to a load profile involves a few Excel tricks. My plan is to share them in subsequent posts. But first, I’ll show you some data from a recent project where I had the opportunity to compare the steam consumption projected by the alarm clock technique with the consumption based on other measurements.  The results were gratifying and affirmed the validity of the approach.

Meanwhile, I hope everyone has a good holiday season.

David Sellers
Senior Engineer – Facility Dynamics Engineering

Click here for a recent index to previous posts