Welcome! Here are the website rules, as well as some tips for using this forum.

# Heat loss calculation accuracy

Member Posts: 10
I would be interested if anyone besides me does a temperature bin load analysis of the previous heating season using actual energy consumption numbers (gas, propane or oil) on new projects to compare to the design heat loss used in the actual design and install of new heating systems?

• Member Posts: 3,537
• Member Posts: 6,506
edited February 2022

Edit: Turns out you can post an Excel SS.

steve
• Member Posts: 21,858
The load calc is really an educated guess, a snapshot in time. Possibly the building is never at that exact condition
Certainly it is a good, necessary starting point, don’t make yourself crazy trying to pinpoint the “number”
Bob "hot rod" Rohr
trainer for Caleffi NA
Living the hydronic dream
• Member Posts: 887
Do not know what last years usage has to do with the installation of a boiler in a heating system.

Find or determine the specific low outdoor temperature (like 0 or less or more.)

Next calculate the heat loss of the buildings envelope envelope and size of the rooms to determines the amount of heating needed to provide the heat in the building (like possibly 70 degrees F.)

Those numbers will determine the size and amount of heating elements needed to heat a building to the desired indoor temperature.

The next step is to add the BTUs needed to account for the heat loss of all the piping installed in the building. that and only that will correctly size the boiler for the building.

Looking at last years fuel consumption does not help the people in the building when a real cold time occurs, like when temperatures get below the norm.

Last years heat load is in the past as that could be in error because last year was a warmer year than normal.

Take for example my sons house in Milford Pa.
House was sized for a hot water heating system, 80,000 BTUs input, the house usually was kept warm and cozy. When the outdoor temperature dropped below 20 degrees the boiler ran continuously, When the temp. dropped to 0 degrees the house dropped to 65, often the temp dropped to -5 the house would drop
to 60 degrees. Fortunately they have a fire place in the house and portable oil filled electric radiators to make up the heat load that is needed,

That boiler is one section to small for the building,
How was the boiler sized, This was a tract home built 25 years ago, Sizing was based on a 15 degree out door low and a fire place for occasional usage to make up the not added BTUs at the boiler.

This an assumption based on a contractors need for installing a smaller boiler, cost less. the tip off was the convectors were installed to accommodate an outdoor temp. of O.

Any other method of sizing a boiler is wrong.

Jake
• Member Posts: 15,457
edited February 2022
@dopey27177
I did several manual J's on my own house and I don't think any of them took infiltration into consideration from what I recall.

The results were my calculated loss was something like 60,000 btu/h @ 0 degrees. At 0F my real loss is higher than calculated, I think 66,000. At -8F my actual loss was 72,000 btu/h. The design temp in my area is 6F but our record low is -19F. My house probably would end up with a 60K boiler or furnace according to the manual J and it would really stink.

I think looking at previous usage and trying to take it into consideration is a very good idea especially if the system was in reasonable working order.

Since the manual J isn't perfect, I feel the more information someone uses to try and set a system up the better.
Single pipe quasi-vapor system. Typical operating pressure 0.14 - 0.43 oz. EcoSteam ES-20 Advanced Control for Residential Steam boilers. Rectorseal Steamaster water treatment
• Member Posts: 3,937
edited February 2022
If you want to design a system for the lowest temperature that you will experience, you will have a system that is oversized 97% of the time which is not a bad thing. The temperature I use for design is 5 degrees colder than the published design temperature for where the house is. The radiators are larger and the loops spaced closer together. Seldom is the boiler larger than it would be for a warmer system.
8.33 lbs./gal. x 60 min./hr. x 20°ΔT = 10,000 BTU's/hour

Two btu per sq ft for degree difference for a slab
• Member Posts: 235
You can use previous years' fuel consumption to figure out the temperature-dependent heat loss (how many BTU/heating-degree-day-hour). I've been very carefully monitoring my setup since November, and I can pretty safely say that my home's heat loss in the middle of winter is 500-600 BTU/HDD-hr.

It is important to use a cold month when ground temperatures have hit steady-state (since you can see heat loss was much lower in the fall/early winter). The "99% design temperature" is 12F for my location, and we've hit 1 day so far with an average temp of 13F, I think, although the temperature at any given hour has gotten down to about 0F. Assuming I wanted to keep my house at a toasty 70F when it was 0F outside, I would need ~600 x 70 = 42kBTU/hr. Now in practice, my boiler got replaced by someone who just used the same size as the previous one, and I have a 140k BTU/hr boiler that so far has yet to fire for more than 25% of the day.

The part that bothers me about using fuel consumption is that it assumes that the boiler is operating at steady-state efficiencies, which, if it's grossly oversized, is probably not very accurate. My boiler is nominally 83% efficient, but in a typical heating cycle it will fire for like 4 minutes, then stop for 5 minutes, fire for 90 seconds, stop for 5 minutes, fire for another 90 seconds, and then shut off when the thermostat stops calling for heat. The numbers I used above are just based on nominal fuel usage, but my boiler is probably not converting 83% of those BTUs into heat in the conditioned space in my house.
• Member Posts: 15,457

If you want to design a system for the lowest temperature that you will experience, you will have a system that is oversized 97% of the time which is not a bad thing. The temperature I use for design is 5 degrees colder than the published design temperature for where the house is. The radiators are larger and the loops spaced closer together. Seldom is the boiler larger than it would be for a warmer system.

True,
But if you design one for design temp you'll still have one that's oversized 96% of the time.

Personally I don't feel it needs to handle the coldest temperatures an area will experience, but I feel design temperature at least in my area is a little too high. That's why I chose 0F even tho the record low is much colder. We see close to 0F or slightly below pretty much every year.
Single pipe quasi-vapor system. Typical operating pressure 0.14 - 0.43 oz. EcoSteam ES-20 Advanced Control for Residential Steam boilers. Rectorseal Steamaster water treatment
• Member Posts: 22,918
I am somewhat reminded of many years ago, when I was teaching Civil Engineering around the time hand-held calculators were just becoming available (personal computers were unheard of...).

One of my minor problems at the time was getting my students to understand the difference between accuracy and precision (most folks think they are the same), but another one was to be able to understand the limitations of their calculations.

The fundamental rule is this: the result of any calculation involving measurements can never be more precise than the least precise measurement in the calculation. Sometimes the precision of a measurement is obvious -- limitations in the graduations of a ruler, for instance. That makes it easy. Sometimes, however, there is variation in the thing being measured, and that will not be evident until one takes multiple repeated measurements so the variability can be determined -- most strength values are this way. Taking repeated measurements and analysing them with statistics makes it possible to not only determine some sort of mean value, but also the probability -- assuming the variation has certain statistical properties -- that the true mean of all measurements is within a certain distance from the calculation and the probability that any particular measurement (or specimen) will lie within a certain distance from the calculated mean.

(parenthetically here, that is the meaning behind such casually tossed around concepts as the "100 year flood" or, here, the 99% or 95% or whatever low temperature).

I don't intend to teach a graduate level statistics course here, however, let's consider two sets of BTU per heating degree day data here -- one being from @fentonc 's chart above, and one -- to which you don't have access -- being the fuel records from Cedric, which go back 20 years. In both one can find a considerable scatter in the values plotted. These values are, themselves, the result of calculations based on measurements, and thus should have some inherent precision. In both cases, however, it is likely that at least some if not most of the scatter in results which is evident is from variation in the objective which is not taken into account in the measurements used: put another way, the heating degree day value used in the calculation is not the only parameter affecting the result.

Can we evaluate this variation without knowing (or caring!) what it is to come up with some reasonable conclusions -- in these cases, how much heating power to heat this building? How likely is it that that estimate will be too high -- or too low? (for the pedantic, the following calculations are based on the assumption that the data used are normally distributed. They aren't -- more like a Poisson distribution -- but for our purposes here that's not critical).

@fentonc makes a very reasonable and completely valid assessment, by eye -- his home's heat loss in midwinter is somewhere between 500 and 600 BTU/Hdd-hr. And, being evidently a prudent chap, selects the higher value to base a conclusion on. There are, however, higher values -- how likely are they to occur? Without number crunching, it's a bit of a guess. In the case of Cedric, however, there is more data, and some number crunching has been done: the average is reported as 57,554 BTU/Hdd-day. But how precise is that value? More number crunching shows that 68% of the individual measurements are between 41,148 BTU/Hdd-day and 73,960. BTU/Hdd-day.

What number should we use for design? Um... err... @fentonc 's approach has the advantage of simpllicity: take a reasonable upper bound and go for it. If you have more information, you can use the statistics above and say, with some confidence, that the actual value of the structure heat loss will be less than 73,960 BTU/Hdd-day 84% of the time. You could go further, and say that 95% of the time the heat loss will be less than 90,365 BTU/Hdd-hr.

Now this isn't a life safety thing, so somewhere between the 84% and 95% values may be reasonable -- and it would be prudent to report the results in a way that didn't imply more precision than you really have -- so one would say a loss of 70,000 BTU/Hdd-day at 84%, and 90,000 BTU/Hdd-day at 95%. and go with 80,000 for your value. (for most life safety things, by the way, engineers will work at 99.9% to 99.999% values).

But surely we have more information than that? Couldn't we report 82,163? No. It looks pretty, but none of the numbers following the 8 have any useful meaning at all. They're just what a really dumb calculator reports. To go back a step, couldn't we just use the overall average -- 57,544 for design? Not prudent --since by definition you would know that half the time it would be too low.

So what of the title of this whole discussion -- heat loss calculation accuracy? May I humble submit if we report a heat loss calculation as a figure plus or minus 10%, which what saying "60,000" or "80,000" -- from the above comments -- we are fooling ourselves. Further, if we base our conclusions on just one, or even a short series of measurements (say a week or two of fuel burn data) there is a very high probability that our result will be significantly off the real value -- but we have no way of knowing which way, nor how much. If we base it off other measurements -- say a meticulous Manual J taking days -- there is probably at least one measurement or parameter in there (commonly the insulation amount and quality) which is similarly limited in precision, which limits the precision of the entire calculation.

I'm not advocating guessing. I am advocating understanding the limitations of what you are reporting.
Br. Jamie, osb
Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England
• Member Posts: 887
What is oversize?

Do you want to drive a car on a super highway with a 80 h.p. engine and that car cannot keep up with the traffic flow or would want that car to have an engine a little larger where you can stay in theslow lane and keep up with the traffic?

In the example I gave at my sons house the boiler was short 1 1/2 sections, so the replacement would be a boiler with 1 more section not two.

The boiler can be a little smaller (little smaller) not a little larger.

We can spend hours arguing over how to size the boiler, remember I said the building envelope must be considered in sizing the boiler. Many boiler installs are boiler replacements some with boilers more than thirty years old (and a third replacement ) where the insulation was removed from the piping (an increased boiler pick up factor). And, many buildings had upgrades such as double and triple glazed windows better store front (building entry) doors, additional roof insulation and pipe insulation.

A proper survey of the building is needed before selecting the proper boiler size is needed. Today we are supposed to be beyond the rule of thumb or guessing about last years fuel consumption. Comfort is the primary thing to supply in most cases the boiler pick up factor will supply the additional BTUs needed because the piping is already hot.

Jake
• Member Posts: 1,796

I’ve found it very useful. I’m blessed enough to have daily gas data and yes, it’s extremely linear and consistent - this graph is about 1200 days. Running the same data using monthly data comes up with a btu/HDD coefficient that’s nearly the same.
• Member Posts: 21,858
ChrisJ said:
@dopey27177 I did several manual J's on my own house and I don't think any of them took infiltration into consideration from what I recall. The results were my calculated loss was something like 60,000 btu/h @ 0 degrees. At 0F my real loss is higher than calculated, I think 66,000. At -8F my actual loss was 72,000 btu/h. The design temp in my area is 6F but our record low is -19F. My house probably would end up with a 60K boiler or furnace according to the manual J and it would really stink. I think looking at previous usage and trying to take it into consideration is a very good idea especially if the system was in reasonable working order. Since the manual J isn't perfect, I feel the more information someone uses to try and set a system up the better.
Lots of versions of load cakes based on manual. The ACCA version gives you choices, some program let you input a number from a blower door test
Bob "hot rod" Rohr
trainer for Caleffi NA
Living the hydronic dream
• Member Posts: 527
edited February 2022

Do not know what last years usage has to do with the installation of a boiler in a heating system.

It is just another tool you can use to make sure you get the right result. If you do the math based on historical fuel consumption and the calculated heat load is significantly different than your traditional heat loss calculation, you should ask why.

e.g. Perhaps the residents like it to be 75F, or perhaps your assumption that all the walls have blown-in insulation is wrong.

• Member Posts: 10
Since I started this post I am very interested in all of the comments. I am a mechanical engineer in upstate NY and have designed and installed numerous hydronic and HVAC systems over the years (too many to note). Being semi -retired, two years ago I designed and installed hydronic heating systems in 3 houses that I could closely monitor and adjust operations in ( a son's, a daughter's, and a close friend's). The house's systems were similar in that they used Mod-con boilers, buffer tanks, a mixture of panel radiators, in slab low temp radiant, higher temp staple up radiant, and some radiant bath walls. The systems were designed to operate on ODR. On each system I started with a detailed heat loss (ASHRAE method) and conceptually used a low system design operating temperature (130 deg F) to maximize the eff of the Modcons and to prepare for the future with a retrofit of an air to water heat pump when the original systems need replacement (corresponding to the necessity of electrification of heating systems down the road). All of the houses were insulated to the requirements of the NYS Building Code, with a mixture of insulating board in the walls and below the basement floor and on site spray foam in the roof systems and sealed tightly. Blower door tests ranged from a low of 0.4 ACH at 50 pascals to a high of 1.8 ACH. (NYS Code requires a Max of 3 ACH)
All of the projects have performed well over the last two heating systems with comfort being utmost in the criteria and operating cost being lower than expected ( all happy owners). That being said I really wanted to critique the systems and my design. To do this I performed a temperature bin analysis on each of the 3 projects using the 30 year temperature bins from upstate NY, the design heat loss analysis, and a fairly accurate record of energy consumption (all 3 cases were utilizing propane). To my surprise the BIN Load Analysis projected energy consumption of from 26-36% higher than what was actually recorded. I have read several articles regarding some implied safety factors in the heat loss calculation and this may be part of it. I also know that aged R values (published values) of Polyisocyanurate foam board are significantly reduced from the R value of the board fresh off the manufacturing line (we tested some early on when it was rated at R8 per inch and it actually tested at R12 per inch). I also realize that things we cannot control such as daily living cycle, the amount of sunshine entering a building, doors and windows being left open, weather conditions varying for the years tested as compared to the 30 year average, that could all have an effect on the analysis.
In the final conclusion I guess a 30% safety factor does provide for a minimum of call backs/complaints and fewer sleepless design temperature nights. For the future I'm not sure that I would automatically reduce the size of the system (bases on the heat loss calculation) but if it was close on a boiler or panel radiator size I don't think I would hesitate to go with the smaller size. With this analysis I have turned the supply water temp to 120 deg F for this season and to date have not had any complaints and hopefully I can see some additional energy savings. I welcome all comments and criticisms.
• Member Posts: 6,506
The calculations @Robert O'Brien noted and used in my spreadsheet are very accurate for residential heating oil, provided the info fed into them is accurate. For gas/propane you'd have to isolate the heater from the other appliances using the same fuel. In my house it was more accurate than the wrightsoft design.
steve