Solving a (self-created) Problem with Heat Distribution

First, some background…

Using a ductless mini-split to heat and cool a home (even a passive house) takes some thought.  One needs to think about where, exactly, the mini-split head, or cassette, will be placed, and how the warm (and cool, in the summer) air that it creates is likely to flow.  Take my home, for example.  We opted to place the “winter” mini-split head in the “living” area on the first floor (we also placed a second, separate, mini-split head at the top of the stairway to the second floor, which is essentially the second floor “hallway,” to cool the entire house in the summer).

That single mini-split head (on the first floor) heats the entire 2,000 square feet of living space (1,000 sqft on each floor) on all but the coldest days, and does the job amazingly well.  When the outdoor temperature is above 20 degrees, the 2nd floor stays within two degrees of the first floor (which, again, is at 70 degrees).  Frankly, that’s warmer than I prefer, since all we do is sleep on the 2nd floor).  But it makes my wife happy.  When it gets below 20 (and the sky is overcast during the day), the 2nd floor can drop to 66 degrees, and if the temperature drops to the single digits for prolonged periods it can eventually go a bit lower. In those instances, I’ll turn on the 2nd floor min-split, and it will periodically kick on, when necessary.

And now, the problem…

When the house plans were drawn, the architect anticipated that the entry to the first-floor office would be directly off of the living area.  He also anticipated that the first-floor mini-split would be mounted on the narrow wall between the living area and the kitchen area.  Both are shown in the drawing below:

(Ther drawing also shows an exterior door on the south wall that I decided against because it just wasn’t necessary.)

When I framed the interior of the house, I decided to move the office entry to the hallway because it unnecessarily ate up wall space in the living area, leaving no room for a T.V., for example.

I also had to find another place for the mini-split because the wall that the architect had intended to place it on was too narrow for the mini-split head; it wouldn’t fit.

Given those issues, I changed the first floor as follows:

Moving the office door to the hallway (and eliminating the unnecessary hall closet) freed up valuable wall space in the living area and widened the hall.

And as the drawing also shows, I moved the the mini-split to a wall that had suffficient room.

Here’s how it looks:

In this photo, you can see the location of the mini-split relative to the back hall:

And here is a similar view in which you can see the door to the office: 

So what’s the problem?  Somewhat ironically (at least in my view) the only problem that we’ve had is that, during the winter, the office, which is the room closest to the mini-split,  can be two to six degrees colder than the rest of the house, depending on the outdoor temperature. The only exception is during sunny days, when the office warms up perfectly.  But that leaves many winter mornings, days, and evenings when my wife isn’t happy (because the office is her place of work).  And it should be noted that she almost always works with the door fully open.

I pondered over this problem for two years.  Initially, I thought that the problem was the ERV.  More specifically, there is only one first-floor ERV supply, and it is in the office.  Why?  Because that’s where Zehnder said it should go.  And it’s consistent with the general rule; pull stale air from the kitchen/baths/laundry room and push fresh air into the bedrooms (the office could also be considered a bedroom).

But I eventually found out that the ERV wasn’t causing the problem, and had no detectable (at least by me) effect on the room temperature.  So while the ERV is constantly pushing cooler air into the room, the temperature (above 60 degrees) and flow ( about 24 cfm) of that air just isn’t enough to make a difference in the room temperature.  This should have been no surprise, as it is consistent with what the experts suggest.

So my current theory is a bit different.  First, after framing the office walls, I insulated them Roxul so my wife could work in relative peace if I chose to watch TV; admittedly a bad idea with regard to heating the room. But I also suspect that any warm air that makes it into the rear hall is probably more likely to rise up the stairs (which are to the right) than into the office (which is to the left).

The easy thing to do would have been to install some baseboard electric resistance heat in the office, which my wife (or a thermostat) could turn on as necessary.  But that clearly seemed like a wasteful overkill to me.  So what I decided to do instead was install a Tjernlund ASI AireShare Room-to-Room Ventilator fan. The idea came to me as I was painting the wall in the family area that separates the family area from the office.  Unsurprisingly, the air near the ceiling (which is almost 9′ high) was several degrees warmer than the general temperature of the house down among the living.  It therefore seemed reasonable to believe that, if I could push some of that air into the office, the problem might be solved and, since the fan only draws 25 watts at full speed and pushes the air at 75 cfm, I figured that its use would prove to be much more user-friendly and cost effective than baseboard heat.

If the intent is to move warm air, ideally, these ventilator fans would be installed with the intake high on one side of a wall and the diffuser low on the opposite side. In doing that, the bay in which the fan and diffuser are installed acts as the vent; air enters from the top and is pushed out through the bottom. However, I decided not to do that because, as I stated earlier, I had insulated the office walls with Roxul, and I didn’t want to completely remove the insulation from the bay in which the fan was to be installed.  Of course, this meant that the warm air would be brought into the room closer to the ceiling than the floor, which is less than ideal.  But the room isn’t that big (it’s about 120 square feet), so I was hoping that it wouldn’t matter.

With minimal effort, I was able to cut the hole on the intake side of the wall about a foot from the ceiling, clear out the insulation in a space of about 18 inches, and seal off the bottom of that space with a piece of 1/2″ foam board and caulk.  Prior to sealing off the bottom of the space, I fished Romex up from the basement and installed a variable speed wall switch. The intake vent and fan are located at the top of the space, and the diffuser is located at the bottom.

You can see the intake for the fan in the upper left-hand corner of this photo…


…and the exhaust above the chalk/cork boards in this photo (it’s a bit difficult to see because it is a narrow diffuser rather than a traditional vent)…

Here’s a better photo of the diffuser (my apologies for the temporary “flow-detector”) and the variable speed switch…

The fan has now been in service for about a month, and I am happy to say that it has performed better than I had anticipated.  So far, it has resolved the issue completely.  For example, last night the low was 23 degrees, and the entire house, including the office, maintained 70 degrees +/- one degree. While not annoying, the fan is detectable at full speed.  However, we run it at what I would estimate to be one-third to one-half speed, and it is essentially silent.

Update on Power Usage – One-Year of Detailed Data

As of December 7th, the eGuage energy monitor has been installed for a full year, giving me the ability to know how much energy it took to heat our house for a full one-year period.  The data was actually a bit better than I anticipated:

Total kWh used to heat the house for the period December 7, 2014 through December 6, 2015:  2,549.6534

At 15 cents per kWh (winter rate) that resulted in a total cost of $382.45 (This includes all charges on our electric bill).

Knowing that natural gas is generally much less expensive than electricity, I thought it might be interesting to compare the two.  In other words, if we heated the house with natural gas, what would the cost have been?  Here are my calculations.  Feel free to let me know if you think my numbers are incorrect:

One kWh of electricity will generate approximately 3.14 kBtus of heat energy.  Therefore, our house used 8,007 kBtus of energy to heat the house during the one-year period.

8,007 kBtus equals 8,007,000 btus.

A “therm” of natural gas produces approximately 100,000 btus and costs approximately 79 cents (of course, this varies by location).

Therefore, if our house had a 100% efficient natural gas furnace, it would have used slightly more than 80 therms of natural gas for heat (8,007,000/100,000) and that would have resulted in a cost of $63.25 for the year.

Now, of course, we can’t use natural gas for several reasons; natural gas is not available where we live, they don’t make a natural gas furnace small enough, and a Passive House is so tight that burning any fuel in it is not advisable.  But I thought the comparison would help put the energy efficiency of a Passive House in perspective for those who use something other than electricity to heat their home.

Some other highlights from the one-year data:

The water heater used 715.516 kWh; about 31 cents per day.

The clothes dryer used 634.7463 kWh; about 28 cents per day on average.

The ERV used 369.02674 kBtus; about 17 cents per day.  However, it should be noted that the ERV was off for an estimated 50% of the three summer months because we had the windows open.

Raw Data on Electric Usage and Temperature/Humidity

In this post, I’m including links to files that contain the historical electrical and temperature/humidity data for our house from February 1, 2015 through the date of this post, October 23, 2015.  The data begins on February 1st because although I started recording in early December 2014, the heat pumps were not working correctly until late January.  My intent is to offer this data to anyone who desires to use it for analytical purposes.

The electrical data, which was recorded by an eGuage energy monitor, contains daily numbers for total electrical use, the first floor heat pump, the second floor heat pump, the combined total of the two heat pumps, the dryer, the water heater, the ERV, the well pump, and the family room, which basically includes the lights and TV in that room.

The temperature and humidity data was recorded by Hobo data loggers that were placed on each floor (basement, first floor and second floor), and outside on the north side of the house under the porch roof.

eGauge Data 2-1-15 thru 7-31-15

I will update this data on February 1st of next year.

In early December (once I have a complete year of eGuage data), I will post the yearly total electrical usage for the appliances and mechanicals that are listed above.  As I indicated, the heat pump data won’t be totally “accurate” (because they were “short cycling” until the end of January).  But the information should still provide a reasonable indication of how efficiently the house is operating.

Energy Usage Update

As I said in the last post, I installed an eGauge energy monitor on December 7th.  That gives me the ability to isolate the energy used by the heat pumps. The two heat pumps have used a total of 899 kWh during the 53 days that have since passed.

Broken down further, the first floor unit used 682 kWh during the period, and the 2nd floor unit used 217 kWh.   I anticipate that this ratio will change significantly in the future because the 2nd floor unit idles most of the time, rather than cycling on and off every three to five minute as it did before the remote thermostats were installed.

So overall, that’s an average of just under 17 kWh/day.  At 17 cents/kWh, that amounts to $2.88/day.

During the period, the thermostat for the first floor has been set at 70 degrees, and the second floor thermostat has been set at 68 degrees for the most part.   According to the electric company, the average daily temperature was 35 degrees for the 35 days ending January 22nd.  The average temperature for the month prior to that was 41 degrees.

One side note…I spoke with the electric company after receiving the January bill, and asked if they had a special rate for homes with electric heat.  They do, and will be reducing my rate to 15 cents/kWh on my next bill.

Short-Cycling Mini-Splits

On December 7th, I installed an egauge energy monitor on twelve circuits; the two lines supplying power to the house (2), the two heat pumps (4), the water heater (2), the ERV (1), the well pump (1), the family room circuits (1), and the dryer (1).

At about the same time, I set the first floor mini-split at 70 degrees and the second floor unit at 66 degrees.

One of the first things that I noticed when reviewing the egauge data, was that the heat pumps were both “short-cycling;” that is, they were turning on and off at intervals as short as three to five minutes.  When “on,” the unit would draw about 1kW.  When “off,” the unit would draw about 20 watts; just enough to power the internal fan on the wall unit (which runs continuously).  Here are a couple of examples:

Screen Shot 2014-12-08 at 1.12.28 PM


Screen Shot 2014-12-09 at 1.57.59 PM

This didn’t always happen.  Occasionally, one heat pump or the other would “flat line” at 300-400 watts.  This could go on for hours, and in a couple of instances, days.  But then the unit would revert back to the short-cycling.

Screen Shot 2014-12-13 at 3.51.27 PM

The short cycling didn’t seem to have a strong correlation to outdoor temperature.  For instance, sometimes the first floor unit would short-cycle while the outdoor temperature was in the 30s and flatline when it was in the 40s.  At other times, just the opposite would occur.

Having said that, the first floor unit stopped short-cycling when the temperatures dropped below 20 degrees or so.  But the 2nd floor unit continued to short-cycle even at those low temperatures.

I contacted both the local HVAC company that installed the units and Mitsubishi’s National Sales Support Manager John Bart, and presented both with a summary of the data.  I had learned about John when we were beginning construction of the house.  He had participated in a Passive House webinar where he discussed the Mitsubishi mini-splits. Shortly after that webinar, I contacted him with questions about the type of mini-split system that might be best for my house, and he was very responsive.  In fact, it was his responsiveness that caused me to go with Mitsubishi over the less expensive Fujitsu units.

It appeared to me that the local HVAC guys (who installed the system) didn’t know quite what to make of the data.  However, John, who was clearly more familiar with the Passive House concept, asked for additional information; a floor plan, photos of the wall units, and the Passive House heat load computation, which was (is) about 8,000 btu/hr.  Given that information, he surmised that the units were oversized, and sent the information over to his Northeast Tech office.

To back up, both mini-splits are 12,000 btu units, and can power down to somewhere around 3,500 btus.  John suggested that, at the more moderate temperatures, even 3,500 btus was probably too much for the house.  However, he acknowledged that it was perplexing that the short-cycling stopped for sometimes lengthy periods of time, even at those more moderate temperatures.  He added that even the smallest (9,000 btu) mini-splits would probably have had the same issue because their low end was very close to the low end on the 12,000 btu units.  He suggested that a Mitsubishi regional tech visit my house to assess the situation and to install remote thermostats for both units.

Mini-splits (at least Mitsubishi mini-splits) come with a remote control.  However, contrary to what some may believe (in fact, even the local HVAC contractor believed it), the remote does not have a built-in thermostat.  In fact, the thermostat is located in the “cassette” that hangs on the wall and dispenses the heat.  As I understand it, the sensor measures the incoming air, and in doing so, provides the data that helps the unit decide when to power up or down.  But apparently it isn’t like the old days, when a single thermostat turned the furnace on or off.  Rather, as I understand it, with these units there are five sensors; three on the outdoor condenser and two on the indoor wall unit, that collectively tell the unit what to do.  In other words, things are much more complex than they used to be.

At any rate, last Friday the Mitsubishi tech showed up and installed the two remote thermostats.  These thermostats (product MHK1) are manufactured by Honeywell specifically for the Mitsubishi mini-splits, and it appears that they resolved the issue.  In my opinion, Mitsubishi provided outstanding customer service in dealing with this issue; something that I find sorely lacking with many companies these days.

Since last Friday, the 2nd floor unit has been idling; meaning that the compressor has not turned on.  In other words, even during the evening lows, which have gone down to the low 20s, the first-floor unit has handled the entire job of heating the house, and the second floor has not gone below 68 degrees.  So the picture, from the egauge perspective now looks like this for both heat pumps:

Screen Shot 2015-01-20 at 4.10.23 PM

A couple of notes…The second floor heat pump is the horizontal red line at the bottom of the graph.  It’s showing about 30 watts.  As I said earlier, this heat pump has been in idle mode since the remote thermostat was installed.  The more prominent red line is the first floor heat pump.  The outdoor temperature early this morning was in the high 20’s, so my guess is that the three vertical jagged lines are defrost modes.  However, they’re the only three such lines that occurred during that night.  So I could be wrong.  They may instead be cycling blips.  But at one hour intervals, they’re a far cry from the three to five minute intervals that were occurring.