Posted By Dana1 on 11 Mar 2013 11:56 AM

Posted By docjenser on 08 Mar 2013 01:24 AM
Indeed I used the indoor conditioned temp of 72F. The reason for that is the fact that it took 1056 gallons to heat the house up to 72F, not 65F. If you are going of usage data, the internal and solar gains are already included and accounted for. It took an additional 1056 gallons (in addition to internal gains) to get to 72F, not 65F. You don't have the internal gain BTUs to add to the ones you put in the house, they are already included.

That's just bad modeling.

If you're using 65F as the zero point for one linear interpolation (the HDD base), then move it to 72F, you've completely skewed the model.  It didn't take an 1056 gallons to get to 72F- it took 1056 gallons to reach the base-temp balance point where heat load=0 with an interior temp of 72F. That temp in most houses will be within a degree or so of 65F, which is why it's become the standard base temp for heating & cooling energy use in single-family houses.  (In larger mult-unit buildings it's usually more appropriate to use 55F as a base temp.)

Heat loads are approximately linear below the balance point (the zero-load outdoor temp) which is where the internal gains are autmotically included.   If the presumption is that the loads grow linearly below 72F you are explicitly NOT including the internal gains, and setting the zero at 72F.  This is not an insignificant error when projected down to -6F.  Using 72F as the zero point is correct if doing a Manual-J that also accounts for the internal gains as separate entries in the model, but is just bonkers when deriving it from fuel use against 65F HDD. Use a different base temp if you think it's dramatically different from 65F, if you like, but don't mix zero points in the mode or you will get paradoxical results and a crossover point.

The linearity is only approximate, since the R-value of the pink fluff in the walls and attic changes with outdoor temp. In a -6F attic the effective R-value will be slightly lower it's rated performance, but in the walls it will be higher.

Bad modelling...eh? What you are doing is shifting the curve parallel, I am shifting the steepness of the curve. If you live in a theoretical world, you think that every day the temp drops below 65 at night the house will need heat, despite a high solar gain in the summer during the day. If you look in the HDD table below, you have hundreds of heating degree days in May, June, July August, and September when sometimes the temp drops below 65F degrees and is accounted for in the HDD but in reality the house is still heated up from the solar gain and does not use a single gallon. In reality the heat is not turned on between Jun1st and Oct 1st, so it reduces the HDD when the house is actually heated, and it increases the BTUs/HDD in your math. Since the heatloss is not linear in the real world (more like a saturation curve) and only linear at colder temps (when the house does not benefit from the daily solar gain). So increasing the steepness of of the curve actually represents a better real life scenario. Description: Fahrenheit-based heating degree days for a base temperature of 65F
Source: www.degreedays.net (using temperature data from www.wunderground.com)
Station: Caledon, Caledon, ONTARIO, CANADA (79.99W,43.91N) Station ID: IONTARIO339
Month starting HDD
12/1/11 1122
1/1/12 1263
2/1/12 1098
3/1/12 747
4/1/12 745
5/1/12 262
6/1/12 163
7/1/12 62
8/1/12 103
9/1/12 306
10/1/12 574
11/1/12 919
Total 7364

Posted By Dana1 on 11 Mar 2013 12:21 PM

Posted By docjenser on 10 Mar 2013 11:10 PM
The problem with using the actual usage data is that it has no safety margins anymore, like a manual J would have. So the 85% efficiency is my safety margin. It also does not forgive that the homeowner forgot about one of the fuel deliveries, or has kept his thermostat at 68F and now puts it at 72F. Or his successor in the house now keeps it at 72F, or 74F. So common sense is always needed as a reality check.
So a 4200 sqf house, 32 years old, in 7900 heating degree climate with -6F design temp is not a 3 ton house. 4 ton I can see if the house is very efficient, supported by the usage data.
Glad to entertain Curt a bit!

There is inherent margin built into the assumptions that the oil burner really runs at it's steady-state thermal efficiency while burning the entire 1056 gallons of fuel, which it very clearly does not.  As-installed, oil boilers are typically 3x oversized or more for houses that size, and anything over 1.7x oversizing pushes it a bit over the left side cliff on the regression curves.  With typical dumb-control high mass boilers, even at 2x oversizing the won't quite make 80% efficiency, and ~78% would be a more appropriate "as used" AFUE to assign for a mid-80s AFUE boiler, according to people who know how to measure & model stuff correctly.  (< see table 3.  The regression curves for each system can be found in the appendices at the end, if you want to estimate what 5x oversizing does.)

Hot air furnaces have somewhat different, flatter curves on the furnaces themselves, but unlike hydronic boilers they take huge efficiency hits on distribution, from duct leakage and air-handler driven infiltration factors. Even with reasonably sealed & insulated ducts those factors add up to ~10% of the total, and in ducts with out mastic (or FSK tape in good condition) sealed duct joints and seams, with NO duct insulation, the hit is 15%+.  (I can dig up credible studies for those numbers too, if you like.  A lot of good work was behind duct-leakage and oversizing limits in the California Title 24 regulations.)

Common sense? Really? Common sense says that you pay the big bucks to get the efficiency for how YOU are going to run it, not your successor.  If the next owner wants to keep it 78F in there, let them! They may pay for the additional power used in the auxilliary heating, but it's insane folly to buy the future owner and extra ton of GSHP on the theory that they just MIGHT run it a lot warmer than you do. (In my book even 72F feels too warm for comfort in a tight house with 30-35%RH interior humidity, but mi esposa begs to differ. :-) )

I could counter that notion with the fact that also heatpumps do not have their steady state performance when cycling on and off. No question the 85% assumption is part of the safety margin, but so is the real world assumption for different homeowner preference when the house gets sold. I simply do not have my sticker on a system which is designed to less than 70F in our climate. Homeowners change and grow older, and typically raise the temp higher with age. If you have been doing this for a while you will see that a lot, especially when people come off propane or oil.
They run the thermostat low on propane to save money, then the raise it slightly, and then more and more when they enjoy the comfort of the systems. And the least I want is a new homeowner running around and telling people that the geosystem we have installed runs soooo inefficient with the supplemental heat turning on soooo much, just because we undersized it for the previous homeowner.
I am strictly against over sizing, but if you go out there and actually monitor your systems, you know what it takes to keep your customers happy (even your future ones), make the systems run efficient and install them at a low as possible price. None of this has something to do with sitting in front of a computer screen and telling designers and installers what they need to do to act in the interest of their customers. What you call an "insane folly" I call taking care of our reputation. A system oversized for the 68F where you might like it might be just right for someone who runs it at 72F (delivering 97% of the load).

"But everything is subject to gross errors if you're asleep at the wheel. Fuel-use analysis methods fail pretty badly when the occupants had set the temp way back while they took a mid-winter 2-week trip to the Amazon or Aspen, and neglect to mention that teeny little fact. "

Don't I know it. Had customers who turned the furnace to 50 at night. When originally asked they "keep the thermostat at 68".

Yes, inserting a 7F offset into a linear approximation where the linear scaling factor had been derived from a zero point at 65F is complete BS from a modeling point of view- there is no good rationale for it.

Inserting an offset is just inserting yet another error into the approximation, and does neither corrects nor compensates for the non-linear aspects of the true curve in any meaningful way. If you're making a linear approximation, be consistent about where the zero point is for that approximation, and make the non-linear aspects argument on a rational basis, not an offset fudge-factor.

Introducing the offset doesn't increase the steepness of the curve, it's an offset change, not change of slope. The BTU/degree-hour number is still derived at base-65F- you're just moving the zero up by 7F when making the load calc.

Using the nameplate DOE efficiency rather than a modeled system efficiency in fact DOES increase the slope, and provides sufficient margin in any reasonably monitored/reported fuel-use calc. The name plate efficiency is an upper bound that is rarely achieve in real-world systems, and the efficiency at the shoulder seasons is WAY below the mid-winter efficiency. As I've stated before,mid-winter billing periods are far more accurate than annual numbers due to higher duty cycle and a lower fraction of standby. With old hydronic boilers kept at idle all summer to avoid condensation & leakage that summertime standby loss is a huge skew-factor that needs to be dialed-out, since it steepens the slope of the linearization too much, and it's not simple to model (though the modeling done by Brookhaven Nat'l Labs' can get you pretty close there, if you have more info on the exact boiler used.)

Yes factors like solar gain, changes in U-factor on wall assemblies with temperature, a VERY significant improvement in window U-factors at the temperature extremes if there are any low-E coating on any surface of the windows, burner efficiency relative to duty cycle, distribution losses, etc. Even with clear glass the marginal thermal performance of this house gets somewhat better, not worse at low temperatures. With low-E glass it's better still, since the performance of the E-coating is a function of the differences in fourth power of the absolute temperatures (relative to absolute zero) on each side, yet the labeled U-factor is at a single nominal temperatures. (The tested & modeled temperature standards used for U-factor labeling of windows are different in European and the US, which means you can't just make the BTU/hr per square foot per degree F to Watts per degree C per square meter unit conversions to compare European and US window performance.

Add it all up, in general for most fiber-insulated houses the BTU/degree-hour improves with every marginal degree- it's a smaller number at the cool temperature extremes than at middle or low-load ranges due to the performance improvement of the fiber in the walls at higher delta-Ts. (The amount varies with fiber type and density.) The models get stickier when foam sheathing is added, since the some foams lose performance with lower temp while others gain performance (and by significant amounts) but that's not in the stackup here. (One needs to derate an inch of exterior iso from a labeled R6 to R5.2 @ 0F, while you can uprate an inch of Type-II EPS from R4.2 to R4.7 @ 0F, etc.) Manual-J and I=B=R do not adjust for these nonlinearities, but better modeling tools can. From a heating designer's point of view it's an error in low single-digit percentages, a smaller error than the typical WAG on ventilation/infiltration air vs. the real rates. But the direction of the error still matters, and in the case of this house favors lowering, not steepening the slope of the linear approximation.

Barring 50F overnight setbacks, linear approximations using fuel use against base-65F HDD and the nameplate efficiency of the fossil burner is typically an upper bound on the true load, but can be pretty close to reality on hydronic systems where the burner isn't oversized. On my own home using the docjenser-approved offset from a base-65F linearization I should have been pretty damned cold when it hit -8F (13F below the 99% outside design temp) since the heat load calculated with the offset would have left me well short of radiation output at the water temps I'm running, but it didn't lose ground. Using 65F as the zero for both the linearization factor and the heat load calc I still had some margin. Monthly fuel billing over the past 5 years here is very consistent in BTU/HDD with 65F as the base temp except for the very late/early shoulder season months.



Even for the aging thinning skin set, only the the insomniacs are likely to be up & about raving at TV cable news during the 99% outside condition anyway, eh? ;-) If you design for 72F at the 99% condition, it'll still keep up with 76F at the 97.5% condition.

Any way you want to model it, the original contractor's estimate on heat load seems way too high compared to typical homes of that size & construction, and they clearly need to dig down on both the real as-is heat load, and where it could be with judicious envelope improvements.