Now for the fun stuff; an experiment you can perform at home (for your kids science project maybe?) that demonstrates some of the stuff I have been discussing in the previous posts and also proves that a watched pot does in fact boil.
The idea behind this experiment is to follow what happens to water and the vapor over the water surface in terms of temperature as you apply heat to the water. To do that, I took an old tea kettle that we had, put some water in it and then inserted probes from a data logger into the water and into the air above the water.
The reason I used an “old” tea kettle was that I intended to totally boil off all of the water and then (briefly – with out setting the house on fire) superheat the steam. Meaning that at some point, the empty metal container was going to be sitting on a red hot burner and as a result, could become discolored.
Because of the physics of saturated systems, which I have discussed in previous posts, my expectation was that as the water approached saturation and started to boil, the air above the water would also become saturated and then be displaced by pure water vapor. And once the water had totally boiled off, the temperature of the water vapor would rise pretty quickly.
The probes associated with the data logger that I was going to use to track all of this were pretty massive relative to the thermocouple that I have for use with my Fluke multi-meter. The mass means that (all other things being equal) the data logger probes will take longer to respond to a temperature change than the less massive Fluke thermocouple.
But, the limitation on using the Fluke and its less massive thermocouple to monitor the entire experiment was that if I wanted to log quickly (which I felt I would need to do to capture the superheating event) then I would run out of memory. Specifically, logging at once a second, the Hobo logger would last several hours, but the Fluke logger would only last a minute or two.
My plan was to delay the start of logging with the Fluke until the point where I had just about boiled off all of the water. That would let the Fluke capture the superheat event while the hobo logger captured what happened prior to that point in time.
Here is a picture of the tea kettle with the problems in place right before I started the experiment. As you can see, I used a probe from the data logger along with the Fluke probe in the air above the water and a second probe from the data logger in the water that I would boil.
Some of you may have read a post I did a while back on relative accuracy. The point was that real sensors have tolerances with regard to accuracy, not matter how well they are manufactured. As a result, two temperature sensors selected at random will probably not read the exact same value when subject to the same condition even though they would be deemed “accurate” relative to their specifications. Thus, if you are comparing the readings, they would show a temperature difference that didn’t actually exist.
Since a lot of the operational decisions and observations we make for HVAC processes (including the ones we are about to make with regard to the tea kettle experiment) are made based on differences in parameters like temperature and pressure, its important that we eliminate the false differences created by the sensor accuracy issues before using the data.
My approach for doing this when I am data logging is to log a string of data with the sensors all subject to the same stable condition before deploying the logger. In this case, I twisted the three sensors I was using together and tucked them inside a box of Kleenex that was sitting on the kitchen counter and logged data while I set up the rest of the experiment.
Once the sensors had stabilized inside the Kleenex box, I would have a string of data for all of them that should be exactly the same. After the experiment, when I had exported the data to a spreadsheet to work with it, I could average the data for each sensor over the stable period and use that value as an offset to correct two of the sensors relative to the third.
As you will recall from my previous discussion of how saturated systems work, prior to saturation and subsequent to saturation, the addition of energy to the system will cause the temperature of the fluid to increase. But during the transition from a liquid to a vapor, the addition of energy to the system will not cause the temperature to change since it is being used to change the fluid from a liquid to a vapor.
That meant that I would need to have some way to monitor the power into the stove. One way to do this would be to install CT on the wiring to the stove and measure the amps as the experiment progressed. But in my case (being a true nerd) I actually have a Brultech power monitor on my electrical panel that is currently monitoring:
- Incoming power;
- Power to my shop and Kathy’s art studio;
- Power to our water heater, which is electric;
- Power to our washer and dryer which are also electric; (Kathy and I live in the Pacific Northwest where 64% of the power is hydro-electric, so the carbon footprint and source energy foot print of electric heat is a bit different from someplace like Indiana, where the majority of the power is generated by coal fired plants, based on the 2004 data from eGRID)
- Power to my office;
Here is a picture of the actual monitor …
… and the CTs on the incoming line.
Bottom line was that if I coordinated my experiment to happen when there was nobody in the shop, office, or art studio and when the water heater was off line and when we were not doing laundry, then for all intents and purposes, the change in consumption I would see would be the stove. So, I could subtract the average baseline before I started from the power consumption readings during the test to identify the power used by the stove.
Watching the Pot
At this point, all was ready to run the experiment. All I needed to do was watch the pot, a task I participated in along with my trusty Nikon CoolPix camera. I pretty much keep it on my belt. I think digital cameras may be on of the greatest engineering tools ever invented for capturing field issues.
Plus you often seem some pretty cool, funny, or memorable things …
… and being able to reach back, pull it out and capture the moment has been a true blessing on many an occasion.
In any case, my camera on a tripod allowed me to monitor the progress of the experiment from the point where I first turned on the burner under the tea kettle …
… through the gradual increase in water temperature (keep your eye on the Fluke, which is monitoring but not logging at this point, and the clock on the stove, and bear in mind that the Fluke is reading a thermocouple, so it could be off a degree or so) …
… to the point where there started to be a little bit of noise from the pot (meaning bubbles of steam were starting to form on the bottom) …
… to the point where the “rumbling” noise was pretty steady …
… to the point where wisps (a technical term) of vapor were starting to show up at the spout of the kettle …
… to full fledged (another technical term) boiling. This, of course, is tangible evidence that a watched pot will actually boil.
The boiling lasted for a couple of minutes until all of the water boiled off (I started the Fluke logging process about 30 seconds before taking this picture) …
… which led to superheated steam (and a house on fire if I didn’t shut down things pretty quickly).
To quote sheriff Andy (0r maybe it was Gomer Pyle) (either way, you probably have t o be about my age to get that) “Excitement, excitement, excitement!”
Taking a Look at the Data
Here is the data from the experiment.
The purple line is the incoming power to the house and the big jump is where the stove was turned on. You read it against the right axis.
The other lines are all temperatures and are read against the left axis. The red line is the temperature reported by the data logger probe that was in the liquid water.
The dashed green line is the temperature reported by the data logger probe that was in the space over the liquid water but inside the tea kettle. Initially, this was a mix of air and water vapor, but as the water boiled, it likely because pure water vapor since the steam produced by the boiler was somewhat contained by the lid on the pot and I suspect forced the air out the spout.
The solid blue line is the temperature reported by the Fluke thermocouple. You can see what a difference mass makes in terms of sensor response by comparing the red, green and blue lines right before 7 pm, when all of the water had boiled off and the steam that remained was being superheated.
As you might expect the mass introduces a lag in to the temperature sensing process because the sensor it self has to be heated up by the energy in the stream you are measuring. For a sensor in an operating HVAC systems, the lag becomes part of the apparent dead time – the difference in time between when a change actually happens and the control system “sees” it. Lags are generally the enemy of tight control so minimizing them can be important. In one instance, I actually took a sensor out of a thermowell and installed it directly in the process stream to eliminate the lag created by the thermowell.
While I did not like the fact that this meant I would have to shut down and drain the system to change out the sensor, it eliminated a lag and helped me minimize the deviation from set point and settling time when the process was upset. And, unlike some of the other lags in the process (controller processing time, transportation delays, the heat transfer characteristics, etc.), it was something I could get my hands on and eliminate fairly easily.
Step 1 – Heating the Water to Saturation
Returning to the data, notice how the power input through-out the process was relatively steady.
Initially, the steady power input causes the water temperature to rise.
If you do the math on the energy during this period, you discover that the stove is about 50% efficient in terms of heating the water. I other words, for the 2 cups of water that I was heating, it should have taken about 40 watt-hours to get it to the saturation temperature and it took a little over 80. No wonder the kitchen gets hot when you cook (or do science experiments).
Of course, starting cold, the stove also had to heat up itself (i.e. the metal in and around the burner) and the tea kettle. So, some of that energy was tied up in the mass of the stove and the tea kettle.
Step 2 – Converting the Water to Steam
The real point here is that for while the steady input of power caused the water temperature to rise initially, things change when the water temperature reaches about 212°F, which happens to be the saturation temperature for water at atmospheric pressure. As you can see from the graph, once the water in the tea kettle reaches the saturation temperature, the steady input of energy no longer causes a temperature increase.
That’s because the energy going into converting the liquid water to steam, a process that requires about 971 Btu/lb at atmospheric pressure instead the 1 Btu/lb-°F that it took to raise the temperature of the water.
In our experiment, that means it takes much longer to change the two cups (about 1 pound) of water to steam as compared to what it took to heat it up from 80 to 212°F(about 296 watt-hours vs. the 40 watt-hours). Instead of increasing the temperature of the water/vapor mix, the energy is going into making the molecules move faster and be able to stay farther a part, transforming the water from a liquid to a vapor.
Step 3 – Superheating the Steam
Once all of the water has been converted to vapor, the temperature starts to rise again, very quickly.
Once again, the energy input is mostly causing the temperature to change (all though it is impacting the internal energy of the molecules a bit too). And the temperature rises much more quickly than it did with the steady application of energy to the liquid water because it only takes about half a Btu/lb/°F to change the temperature vs. the 1 Bt/lb/°F associated with the water.
The other factor that comes into play with regard to the sharp temperature rise is that when we first started, there was about a pound of water inside the tea kettle. But when we converted the water to steam, it went through a tremendous volume change, meaning there is only a fraction of a pound of water vapor contained by the kettle.
This screen shot from the saturation table I generated for water using REFPROP illustrates this.
The bottom line is that there are not many pounds of steam inside the tea kettle to heat once all of the water has been converted to vapor. Thus the quickly sky-rocketing temperatures.
Hopefully, my little foray into how saturated systems work with the experiment at the end gives you a better feel for this important phenomenon. Having such an understanding along with a “gut level” feel for it will be very useful to you as you work in the HVAC industry and also will help you answer questions from your kids and grandkids about clouds, ice spikes, and other things they might observe in the world around us.
Senior Engineer – Facility Dynamics Engineering