I've been thinking about this further.  Some thoughts:

First:

Dewayne, I'm not sure you can eliminate the effects of solar gain from the experiment, regardless of how much you narrow down the data collection period.  Your house is so tight and well insulated that I believe solar gain is a significant piece of the 24 hour BTU needed for your home.  I believe solar gain equally affects my home, but, because it's so leaky and poorly insulated, it's not a significant portion of the total amount of BTU needed for my home.  Thus I believe I can get away with ignoring solar gain, but you may not be able to.

Your new data will be interesting to look at, to either prove or disprove this thinking.

Second:

Curt, you asked a while back if 65° was the true 'zero BTU' need point for my structure.  Or, in other words, what is the intercept?

Answer: If I collect a lot of data and look at it over time, indeed there's a clear linear relationship between HDD and KWH.  In fact, a year ago, for my home, the relationship is represented as:

KWH = ( 2.2 * HDD ) - 11

See graph below.

For example, if HDD is 5 (60° outside avg), then KWH is 0.  Iff HDD is 30 (outside avg is 35°), then KWH is 54.

Third:
 
My 'Index' for this graphed time period is 2.2, different that the 1.4 Index I came up with last week.  Some guesses as to why might include:

1.) I was using a different setback-recovery scheme.

2.) I added insulation / reduced 'leakiness.'

3.) Indeed a setback-recovery scheme isn't as cost effective as a constant temp scheme.

We'll see from the data if it's possible to make a conclusion.

Last:

Looking at the data from a year ago, I think the experiement is fundamentally flawed if too few days are included in an experiment.  It's clear from the graph that there's a linear relationship between KWH and HDD, at least for my structure.  But, the day to day variation can be significant.  Perhaps Dewayne's experiement is actually just fine on a 24 hour basis, and that it's simply not a long enough experiment period.

And perhaps I can get away with a shorter experiment period because my house is not so energy efficient.  We'll see.

Interesting subject.  I'll have my first day's setback-recovery index number to report tomorrow.

Best regards,

Bill

Bill,  this is certainly an interesting experiment we are doing.
 
My thoughts are thus:

Even though my house significant solar gains,  the room temp  is back to 70° by 9 pm.  This is when my tsat sets back.  If there is no residual solar gain in the house after 9 pm,  (the sun has been down for 4 hours  by then at my place),  then what ever heat my house loses from 9 pm to 4 am  ( when the set back ends)  will not be affected by the solar gain and  seems like a good span to study set back.

This is based solely on my own intuition.

The results will certainly be interesting.

I did set the time on my WEL ahead 3 hours and now the daily data resets at 9 pm so the charts will reflect the data we are looking for.

So from 9 pm last night till 9:30 this morning ( the time that my heat pump shut off after recovery from the night set back) my KWH/HDD was .974.

We'll see what happens tonight.

My KWH/HDD for Sunday Night was .884

This whole KW/HDD question has me wondering about some things.

If there were two identical houses, one in an area where the low temp never went below 40° F and another in an area where the low temp was down to 0° F often, assuming that these two houses are identical in every way except location,  would the KW/HDD be the same for each house?


I guess what I am wondering is, does the KW/HDD go up as the outside temp goes down?

If the delta T across the wall is greater,  does the heat escape faster?

For my structure, that is not exceptionally energy efficient, the answer is no, within reasonable limits.  See chart below, for data collected over a long time period, with many varying weather patterns, showing that for me there's a clear linear relationship between KWH and HDD, and it's not as a function of outside temperature (within reason).

I believe the index (slope) is a function of the structure itself, and the tstat setting(s).

Best regards,

Bill

Hey Guys,
My measurements are just approximations, but you might find them interesting. Just read my meter for the month of January and used about 1,300 KWH because of my Geo furnace. (this is my first winter with it). I subtracted the KWH from other uses in the house by looking at my average usage for the last 2 years. The nearest weather station 50 miles south of me in Green Bay, WI shows 1773 HDD for January. For my 1900 sq ft old farm house that would be .73 KWH/HDD. We were down to -22° for a couple days and didn't get above 32° for the entire month. I don't set back my thermostat at all and my auxillary electric heat strip is turned off by means of the dip switch in the waterfurnace.

Two questions , different answers. In reverse order:

The U-value (1/R) is a measure of the thermal conductivity of the wall structures, and the heat loss is directly proportional to the temperature difference and the surface area.  Bigger delta-T means faster heat loss, requiring more kwh to make up the difference.

The kwh/hdd is more complicated though- there are several factors involved.  The efficiency of the heat pump system isn't the same at every heat load- short cycles under light loads aren't very efficient, nor are very long cycles that super-chill the loops, or when it can't keep up, the resistance heating kicks in.   Furthermore, the hdd number is based only on air temperature, but factors like solar gain through the windows (or lack thereof) dramatically affect the net losses through the walls.  Similarly, wind-induced infiltration varies widely independent of air temperature, and  wind induced turbulence of the boundary layer on the outside of the walls increases the effective U-value measureably, all of which affects how many kwh it takes to keep the indoor temp steady. 

Thermal conductivity ignores radiant & convected heat transfer, and isn't a constant- the R-value of materials can and does vary with delta-T,  some more than others.  (eg: Low density blown fiberglass in horizontal installations like attics has about half the R value at an 80F delta-T that it has at a 30F delta-T, something the cellulose vendors like to point out.  Similar installations of cellulose increase in R value very slightly with increasing delta-T.)  The actual construction of the "identical houses" and heating systems in the thought experment may lead to somewhat paradoxical or at the very least, quite non-linear results.  This stuff isn't easy to model accurately- which is why measuring stuff counts!

As others have noted we have some fundamental measurement issues in this experiment.  When data is taken over only a few days or tens of days the "noise" from all the other factors interfere with the measurement.  It takes 100s of days for the other factors to all average out to give you anything close to 2-digit precision on what the setback is/isn't doing for you.  HDD by itself is a very crude measure of the heat load- it may be the most important factor for most structures, but by no means the only factor for any structure- it's only a first-order approximation at best.  And KWH/HDD or even KWH/AACL (Actual Average Heat Load) is not a linear function- it system-specific and will vary significantly with where the actual averge heat load is relative to the "sweet spot" on the efficiency curve of the system. (It's guaranteed to be a curve, not a line, and a possibly quite complex curve at that.)

Posted By Dana1 on 02/03/2009 9:02 AM

... The kwh/hdd is ... complicated ... there are several factors involved.

... we have some fundamental measurement issues in this experiment.  When data is taken over only a few days or tens of days the "noise" from all the other factors interfere with the measurement.  It takes 100s of days for the other factors to all average out ...

... HDD by itself is a very crude measure of the heat load- it may be the most important factor for most structures, but by no means the only factor ... it's only a first-order approximation at best.

And KWH/HDD ... is not a linear function ...

Dana1, I thought your posting was particularly helpful and instructional.  Many thanks.  I think I'm seeing many of the things you point out.

In my case, I'm helping to stay in more of a linear part of the Index curve by not running any Aux heat and by staying in 1st stage.  I'm also averaging out the 'noise' faster because my house is not substantially energy efficient.

I think 5 days for an experiment is probably the biggest approximation.

Best regards,

Bill

I'm into Day 9 of the experiment.  Here's an update:

The first 5 days were with the tstats all set to a constant 68° on a 7x24 basis.  The result was an avg index value of 1.4 KWH/HDD, and with each individual day's index within 10% of the avg.  Daily results are again shown below:

Sun     22.7 KWH     17.5 HDD     1.3 Index     48° avg outside temp     173 KBTU consumed     Cloudy

Mon     31.9 KWH     22.8 HDD     1.4 Index     42° avg outside temp     263 KBTU consumed    Cloudy

Tue     44.6 KWH     29.4 HDD     1.5 Index     36° avg outside temp     373 KBTU consumed     Cloudy

Wed     44.4 KWH     32.5 HDD     1.4 Index     33° avg outside temp     376 KBTU consumed     Bright sunshine

Thu     37.9 KWH     25.9 HDD     1.5 Index     39° avg outside temp     311 KBTU consumed     Bright sunshine

At the beginning (12:01 AM) of Day 6 (Fri.) the tstats were set to a 3° setback-recovery mode.  The tstats were all set to a constant 68° during the day, and set back to 65° for night time hours.  Day 6 was used as a transistion day.

Here are the first 3 days of using a 3° setback:

Sat     17.9 KWH     13.5 HDD     1.3 Index     52° avg outside temp     140 KBTU consumed     Hazy

Sun     9.3 KWH     10.5 HDD     0.9 Index     55° avg outside temp     78 KBTU consumed     Bright sunshine

Mon    12.2 KWH     17.0 HDD     0.7 Index     53° avg outside temp     95 KBTU consumed     Bright sunshine

For the first 3 days, total HDD = 40.9, total KWH = 39.4, yielding an average Index = 1.0 KWH/HDD

No change to parenthetical notes written earlier.

I'll continue collecting the data for 2 more days.

Best regards,

Bill

Is this on a NWS (65) or Bill's house (60) HDD?

I think we'll need apples to apples HDD values and similar sunshine to properly compare calculated indexes. It has been warmer these past 3 days.

If using NWS HDD and you get a day of avg temp 60, you'll have no kwh used but 5 HDD thus 0.0 kwh / HDD

Just my thoughts...

Curt

I have consistently been computing HDD based on NWS 65°.

If I have a daily average of 60° (i.e., 5 HDD), then I probably will not consume any KWH.  For my home, 5 HDD approximates where everything becomes non-linear.

So far the warmest it's gotten during the experiment period is an avg of 55° this past Sunday.  It hit a highh of about 72°.

Detailed weather info for the experiment period is available here: http://welserver.com/cgi-bin/plot/WEL0043/SampledOutsideWeather.gif .

Best regards,

Bill

need more info
or at least a comprehensive summary

I have significant results and will post hopefully this weekend.

Best regards,

Bill

I appreciate the efforts of both of you on our collective behalf.
One of the benefits of geo is saving money no matter what so I've always encouraged folks to just be comfortable.
There is a group of former propane burners however that are not comfortable with out dialing down for bed (must miss the weight of their foot thick comforters).
I'd love to advise them they could have their cake and eat it too.
Seems like a computer model of this process ought to be doable. Unfortunately my 14 year old (the house IT guy), doesn't grasp this one, anyone know a college kid looking for a research product?
J

Here is my conclusion to the experiment, designed to answer the question, does a tstat setback-recovery scheme really save money with GSHP units.


First 5 days: Tstats set to constant 68° on a 7x24 basis.  Result was avg index value of 1.4 KWH/HDD, and each individual's day index was within 10% of the avg.  Daily results were:

Sun     22.7 KWH     17.5 HDD     1.3 Index     48° avg outside temp     173 KBTU consumed     Cloudy

Mon     31.9 KWH     22.8 HDD     1.4 Index     42° avg outside temp     263 KBTU consumed    Cloudy

Tue     44.6 KWH     29.4 HDD     1.5 Index     36° avg outside temp     373 KBTU consumed     Cloudy

Wed     44.4 KWH     32.5 HDD     1.4 Index     33° avg outside temp     376 KBTU consumed     Bright sunshine

Thu     37.9 KWH     25.9 HDD     1.5 Index     39° avg outside temp     311 KBTU consumed     Bright sunshine


At the beginning (12:01 AM) of Day 6 (Fri.) tstats were set to a 3° setback-recovery mode.  The tstats were set to a constant 68° during the day, and set back to 65° at night (approximately 10 PM - 6 AM, about 8 hrs.)

Day 6 was used as a transistion day for the structure to acclimate.


Next 5 days using a 3° setback.  Result was avg index value of 1.0 KWH/HDD.   Each individual's day index was not within 10% of the avg., though.  Daily results were:

Sat     17.9 KWH     13.5 HDD     1.3 Index     52° avg outside temp     140 KBTU consumed     Hazy

Sun     9.3 KWH     10.5 HDD     0.9 Index     56° avg outside temp     78 KBTU consumed     Bright sunshine

Mon    12.2 KWH     17.0 HDD     0.7 Index     48° avg outside temp     95 KBTU consumed     Bright sunshine

Tue     17.6 KWH     15.8 HDD     1.1 Index     49° avg outside temp     140 KBTU consumed     Bright sunshine

Wed     23.3 KWH     21.0 HDD     1.1 Index     44° avg outside temp     197 KBTU consumed     Bright sunshine


Conclusions:

1. The index value dropped from 1.4 to 1.0 KWH/HDD.  Thus, I think it's at least reasonable to conclude that a reasonable (few degrees) setback-recovery scheme does not incur increased KWH usage (cost).  In other words, feel free to set the tstat cooler at night (within reason).

2. If you accept short, 5 day experiment time periods, and if you accept the many data point graph I published earlier showing a linear relationship between KWH and HDD, then it's probably reasonable to conclude there's actually a cost savings associated with a tstat setback-recovery scheme, within reason.


Comments:

1. There are many variables that affect the linearity of a structure's KWH consumption vs. HDD.  The variables I observed included solar gain of the structure, wind, eveness of the weather pattern across each 24 hour period, and amount of internally generated heat (i.e., cooking, lighting).

2. Using many more than 5 or 10 data points, it's clear my home has a linear relationship between KWH and HDD, for outside temps of 60° and colder (see previously published graph).  I believe this is because it has enough air leakage to average out influencing variables.

3. My guess is my home's air leakage is average.  And thus the above conclusions may be reasonably applicable to many homes, noting the next comment on capacity.

4. The above experiment results were obtained operating the residence at the lower end of heating capacity.  As a result, no Aux. heat, or even 2nd stage heat, is ever required for my home.  And thus my EWT stays in a range of 2 - 8° below 'down deep earth temp' (68°), resulting in constant favorable efficiencies.  I would have reservation concluding tstat setback-recovery is cost effective/neutral if resulting run times cover a good portion of the day (i.e. the GSHP(s) struggle to catch up each day).


Footnotes:

1.) My tstats are set to a 1.2° heating differential (largest possible), and set to not allow 2nd stage to turn on until 40 min. has elapsed in 1st stage.

2.) My structure is 3400 s.f., all single story, with 2 GSHP units (WaterFurnace Envision 5 and 3 ton units, both dual stage, variable speed).

3.) Window treatments were  opened/closed at same general time points each day.

4.) I don't have any aux heat capability.

5.) The experiment focused on KWH monitoring.  There may or may not be a direct correlation to cost depending on electric provider pricing strategy.  In my case, there is a direct relationship, as I pay the same price/KWH regardless of time of day.  But this isn't always the case for others.


Best regards,

Bill

Joe, let's see what Duane comes up with.  It's possible, depending on his results, you'll be able to advise them that they can 'have their cake and eat it too.'  This is based on my experiment results, and based on reasonableness, as long as they have adequate capacity (GSHP(s) don't have to struggle all day to recover) and a reasonably normal structure with respect to air tightness.

See above report.

Best regards,

Bill

I'm sure Duane's results will be coming soon.

'Cliff's Notes Version' for my results are:

1. ... a setback-recovery scheme does not incur increased (cost) ... .  In other words, feel free to set the tstat cooler at night (within reason).

2. ... it's probably reasonable to conclude there's actually a cost savings associated with a tstat setback-recovery scheme, within reason.

There are many notes and cautions associated with the above conclusions, but, this would be a good summary of my observation.


Best regards,

Bill

This is Great Information, I've been manually turning my Therostat down 4 degrees at night for the last few weeks for 17 hours a day. (12 am to 5 pm). Based on these results, I can justify buying a new programmable thermostat to automate the process. I'm hoping to see a significant improvement in my overall energy usage.

Thanks!

 

For people who have time of day electric pricing, net reduced KWH usage during setback-recovery might not translate to $ savings for the day, if they setback at a low KWH cost and recover at a high price.

Regards, Masoud