Musings on the "Design Day"

Jamie Hall

I have too much time on my hands… so I was musing on out industry's way of using the "design day" temperature in designing systems. I've operated in a number of engineering disciplines over the years, and almost all of them make use of certain design parameters in design. How much load will a bridge take? How much turbulence can an airplane wing take? How many gallons per minute will a certain fire take to suppress? What is the peak flow to this waste water plant?

And so on.

In all those cases, the engineered result is intended to be able to exceed by some margin (sometimes small, sometimes rather large) that design parameter.

In fact, while musing, it occurred to me that the design of heating systems is the one are I've hit over the years where we have a design parameter — either the 99% or 95% design day low temperature — and where it is quite routine to design the system to not be able to meet that parameter, on the principle perhaps the client won't mind if they are cold a few times. No margin at all. If we are using the 95%, we essentially guarantee that there will be a few days every winter when the heating system can't keep up. The odds are a bit better with the 99% value (they're guaranteed to be chilly at least two winters out of three)…

Why? Any thoughts on this? Comments?

pecmsg

eastern isl of long

Design days less than 1%

35 - 45% for most

Balance for the rest.

LRCCBJ

and where it is quite routine to design the system to not be able to meet that parameter, 

Well, all I can say is that I have YET to see a single installation where the boiler is too small. In fact, I had to use a HWH with a 140F limit to actually fall behind the building at approx. the 90% point. I did expect to fall behind slightly due to being about 10F below the requirement on the design day.

Nobody is ever too small on oil. And, you really have to try hard to be too small on NG.

On commercial, this can be completely different as there is money to be saved by designing right to the limit.

hot_rod

Good data, thanks.

I think the amount the temperature falls behind becomes the concern.

High mass systems like large iron systems, radiant slabs, may flywheel through an overnight condition of at or below design temperature slump.

Low mass systems, or leaky buildings will be more of a problem for the occupants comfort level.

It probably comes down to the occupants sensitivity or expectation.

I'd also look at the type of boiler. Sizing a non modulating boiler for the 99% will certainly lead to short cycling, possibly 80% of the heating season.

A over-sized mod con, would be less of a problem for cycling at below design conditions.

So two balls to juggle for the design engineer.

Hot_water_fan

How much load will a bridge take? How much turbulence can an airplane wing take? How many gallons per minute will a certain fire take to suppress? What is the peak flow to this waste water plant?

@Jamie Hall surely someone as intelligent as you can see how different all of these situations are from heating loads? Plus who is advocating for sizing to the load with zero buffer? The issue seems to be: contractor picks boiler and doesn’t do a heat loss, it’s massively oversized, then receives feedback to that effect, then reverse engineers the rationale. That seems way more common than people advocating for 1.2x vs. 1.05x vs. 1.3x heating load.

EdTheHeaterMan

I have a theory on that @Jamie Hall. The probability of having a load calculation that meets the 95% criteria and finding a boiler or furnace that is exactly that size is rare.

Also when performing the calculations for an "accurate" heat loss, everyone who has done heat loss calculation has been taught to round up to the next foot, or round up to the next whole number when the answer is a fraction. So by design, a load calculation will be slightly oversized when it is done according to industry standards.

Now, when you come up with a number like 43,872 BTUh for example, for your two bedroom ranch in New England, and you look at the specifications and find that the net output of the boiler that you want to install is 38,000 for the two section boiler and 52,000 for the three section boiler you have no choice put to oversize that boiler by almost 8000.

Of course the larger project the more accurate your calculations need to be. And when you get up over 1 million BTUs, the steps between one size to the next is a much smaller percentage, but there is always some factor of oversizing built into the way we do calculations for heating systems

However, since all of these load calculation software packages can be computer accurate, if you decided that a wall that was 12 foot 8-3/4” inches long should be entered as 12 foot 8-3/4” inches instead of 13 foot you may find that you can use a size smaller boiler or furnace on perhaps 10% of the jobs that you quote.

fentonc

You just need to consider what the failure mode is, and what kind of costs are incurred. I've designed computer chips where you need to provision sufficient off-chip bandwidth, and we targeted the equivalent of the "99%" case, knowing that "things slowing down occasionally" was a fine failure mode, and wasn't worth doubling the cost of an interface that would almost never be a bottleneck in practice.

EBEBRATT-Ed

Don't forget that heat loss calculations don't take into account people, refrigeration, sun or other equipment in the building that gives off heat.

All the above is taken into account for AC.

And whenever you come up with the final HL # there is seldom a boiler or furnace of that exact size so it's usually a little larger not to mention the much discussed PU factors.

DCContrarian

I've never seen anyone use 95% temperature.

The assumption behind 99% temperature is that excursions beyond that will be small and brief. It's assumed for the most part that the house will have enough heat capacity that occupant comfort won't be affected.

Let's say you have a house with a heating load of 50k BTU/hr at 70F indoor/20F outdoor. Let's say it goes to 19F for an hour. What will the heat loss be? The delta will be 51 instead of 50 so it will be 51/50 of the design load, or 51K BTU/hr. So if the heating system is exactly sized, the house will be 1000 BTU short for that hour. How much will it cool off? I believe houses have a heat capacity typically between 25,000 and 50,000 BTU/degree F, so it will cool somewhere between 0.02 and 0.04 degrees during that hour.

One percent of a year is about 88 hours. What you typically find is that 88 hours breaks down into 44 hours that are one degree below the design number, 22 hours that are two degrees, 11 hours that are three degrees, 5.5 hours that are four degrees, and so on. Even with just-right sizing the heat capacity of the house is going to keep the house reasonably comfortable.

Jamie Hall

Actually the percent probability — just like storm event probabilities ("100 year storm") is just that: a probability. It's not a measure of frequency the usual sense. Without going into a long song and dance routine — the 99% figure works out, over any given heating season (taken as 100 consecutive days) to there being roughly 2 in 3 probability that there will be at least one day (24 hours) with the temperature at or below that figure during that time period. It doesn't say anything about how long — or how far — below the temperature the actual temperature will be (the 95% probability value works out to a virtual certainty)

I completely agree with the comments on practical sizing of boilers! You go with what you can get.

@EdTheHeaterMan made an excellent point with the computer calculation bit — back in the old days, we did round everything up — and this gave a slightly conservative value for the Manual J calculations. With the computer applications, we don't — so the calculations are a little less conservative. This is true in most fields of engineering — not just heating — resulting in things being designed (often unintentionally) to lower performance margins. It's an interesting problem.

I just wanted to get people thinking!

RascalOrnery

I would Franky rather have a system that is efficient most the time and struggles a handful of days out of the year than is a hog most of the time to serve as a safety factor for that "one time".

It's like driving a pickup all the time in case you see something free alongside the road some day, you might be more ahead by just buying what you need and drive a fuel efficient car.

Have some backup heat like electric or propane units. We are way spoiled, generations before us grew up with outhouses in winter and a cook stove for heating. My father in law remembers ice on the inside of the windows.

Now you can buy snowmobiles with heated seats and handles. We tolerate very little inconvenience any more. But I get it, tell the customer that and you won't get the job!

jesmed1

We had an unusual opportunity to measure heat loss in a once-in-70-year event in the Boston area, in February 2023. There was a multi-day polar vortex event with record low temps and persistent high winds that broke records going back to the 1950's and earlier. Daytime temps were in single digits, and went into double-digit negatives overnight, with high wind chill.

Based on measured boiler run times, I found our average heat loss during the event was about 30% higher than our zero-degree "design day" heat loss.

So, interestingly, if our boilers had been sized to a 1.4 multiple of design heat loss per ASHRAE recommendation, they would have been almost perfectly sized to handle even that extremely rare event.

So if you figure that event probability as 1 day in 70 years, that's about 1-in-25,000. Which means the ASHRAE 1.4 multiplier takes that 99% design day and extends it up to something like a 99.996% design day.

https://www.wcvb.com/article/dangerous-cold-feb-4-2023-boston-massachusetts-weather/42712533

LRCCBJ

IMHO it would be a grave error to size a boiler to a 1:25000 event. If certain individuals cannot tolerate a 10F indoor drop under extreme conditions, I have four short letters for them…………which I cannot mention on here.

Jamie Hall

Mind you all. I'm not suggesting that for many situations you shouldn't design for a smaller delta T than the degree day would suggest, at least in some situations. That is a rather complex decision, though, and there are factors besides @RascalOrnery 's comment that back in the old days we were all tougher.

I think we are all familiar with those — things like codes, or just who your client is, or for some facilities (hospitals, nursing homes…?) the considerations may be different.

@jesmed1 comment is good too to think about. (that 1 in 25,000 figure, by the way, would be quoted as about a 99.999% event, statistically).

Rather off topic — my objective, after all, was to get folks to think about what "design day" really means, statistically — is this thought: if you deliberately design to a smaller delta T, don't make the mistake of telling your client that what you have installed will meet the larger delta T or the design day delta T, even if a building inspector will let it go. They may not mind — or you may find yourself having to explain why Aunt Minnie is too cold and why your system doesn't work. Not a good conversation to have…

And I certainly wouldn't advocate great oversize, as we sometimes see. That is wasteful. But sizing to meet the design delta T may not be, or may create only a relatively small capital cost penalty, and — given good control technology — will have a very minor impact on overall system efficiency (particularly any type of system which can modulate, such as mod/cons or many heat pumps — or even properly controlled (pulse width modulated) steam systems!)

Kaos

Another common myth. The 1.4x sizing factor is the MAX recommended oversize.

ManJ is very conservative, equipment sized based on a very accurate model will carry the place on days colder than design day.

This is something to watch as a heat pump sized based on that misconception of 1.4x oversizing will generally spend a lot of the heating season cycling.

Modcons are also not immune to this issue. Modulation helps but there is always that minimum fire.

DCContrarian

"Actually the percent probability — just like storm event probabilities ("100 year storm") is just that: a probability. It's not a measure of frequency the usual sense. Without going into a long song and dance routine — the 99% figure works out, over any given heating season (taken as 100 consecutive days) to there being roughly 2 in 3 probability that there will be at least one day (24 hours) with the temperature at or below that figure during that time period. It doesn't say anything about how long — or how far — below the temperature the actual temperature will be (the 95% probability value works out to a virtual certainty)"

But it's not a percent probability. It's a percentile measurement and it is absolutely a measure of frequency. The National Weather Service measures the temperature on an hourly basis at weather stations all over the country, the 1 percentile temperature is the temperature where historically, 1% of the measurements — or 88 hours a year — have been at or below that temperature.

What it doesn't tell you is how far below that temperature it's likely to get, or how often it gets that cold. The first you can get from all-time records. For the second, the NWS publishes for every weather station how many hours a year on average are at each temperature. So, for example, where I am in Washington, DC, the one percentile temperature is 22F; in an average year we get one hour at 5F and zero hours below that; the all-time low is -15F. (Although curiously, if you take the NWS data and add up all the hours below the design temperature it doesn't add up to 88 hours, I'm still trying to figure that out. They do round to the whole hour which may have something to do with it).

What's confusing is that percentiles are most often used in statistics for things like storm event probabilities, particularly with events that are independent and normally distributed. In that case you'd expect that 68% are within one standard deviation of the average, 95% are within two and 99.7% are within three (I think this may be where Jamie got the 95 percentile temperature). But daily temperatures aren't normally distributed; below is NWS data for Minneapolis. To further add to the confusion, some sources report the 99.7 percentile temperature rather than the 99 percentile temperature as the design temperature.

DCContrarian

I would Franky rather have a system that is efficient most the time and struggles a handful of days out of the year than is a hog most of the time to serve as a safety factor for that "one time".

An important part of this calculation is, what is the penalty for oversizing and what is the penalty for undersizing? For a lot of combustion equipment there isn't much of a penalty for oversizing, especially moderately, I guess equipment that cycles more won't last as long. The penalty for undersizing is tearing it out and replacing it.

For heat pump equipment, the penalty for oversizing is lower efficiency almost every day and shorter equipment life. It's easy to add supplemental heat to most heat pumps — a 5000 W heat coil adds over 15,000 BTU/hr, which should give a big cushion — so the penalty for undersizing is the supplemental heat runs more than it otherwise would.

I'm kind of a nerd about these things so one of the things I like to do is to take the NWS data and heat pump manufacturers performance curves and model hour by hour electricity usage in various climates. One of the things I find is that a slightly undersized heat pump often ends up using less electricity over the course of the heating season. Yes, it uses the supplemental heat more, but that's only for a small number of hours per year. The rest of the heating season it tends to be so much more efficient that it more than makes up for the extra hours of supplemental heat.

The caveat I will offer though is that heat pumps are much more complex when it comes to the sizing. It depends on the house, the climate and the equipment.

ethicalpaul

In the 14 or so places I’ve lived, about 1/2 of them houses, I have never once seen one that ever couldn’t hold the thermostat.

Have any of you?

Jamie Hall

@DCContrarian is technically right about the percentile bit. And statistically it's actually even more complicated… I was just trying to get across a general feel for it.

One other minor item… in a sense that 95/99 stuff is a measure (not that good a one) of the variability of temperature — how often and how far it moves from the annual average. That variability actually seems to be increasing (and climate modelling shows it should) with the result that those design temperatures probably are too high (too warm). But that, folks, is a whole other can of worms!

And yes, @ethicalpaul , I have lived in places where the heating system couldn't hold the indoor temperature up. Of the houses I've lived in, I can think of three which couldn't.

pecmsg

https://forum.heatinghelp.com/discussion/comment/1830249#Comment_1830249

My current house. Built 1922. We renovated the 1st floor 20 years ago; the 2nd floor is original wood lath and plaster. Replaced a coal fired converted to oil 1.25GPH Nozzle Thatcher Gravity with a 100,000 BTU/h WM Boiler. Zoned 1st 2nd floors.

Eastern Isl of long design 5°F. When near that and the NE wind begins to blow the couch gets pulled away from the 2 in the wall radiators, the wood cover by the stair's radiator gets removed and the boiler runs probably 100%. When it's at 0°F or below we maintain 65 inside overnight. Once the sun comes up so does the house. Thats close enough for the 1 or 2 times we've lived here with those conditions.

I remember growing up Summer curtains came down and Winter curtains went up. I'm sure that would help here.

jesmed1

https://forum.heatinghelp.com/discussion/comment/1830218#Comment_1830218

I fully agree. Ironically, our boilers are sized to 3.5x of our heat load, so in fact they're sized for the Last Glacial Period 10,000 years ago, which would be a 1-in-3.65 million event.

Sizing to ASHRAE would still be oversized, but less egregiously so.

Jamie Hall

Oddly enough, global mean temperatures weren't really all that much lower during the glacial maxima. It does take a relatively short term (century or so) significant drop (about 3 Celsius) from the historical (not current) global temperatures to start one, but not to keep one going. Refer to my Doctoral dissertation on the dynamics of continental ice sheets for more murky information…😉

hot_rod

it was the windy days that hit us the hardest in the 1960 vintage home we had in Missouri. That is a tougher load to predict for

I had a blower door test done after the first year we lived there to find and seal the leakage

jesmed1

https://forum.heatinghelp.com/discussion/comment/1830310#Comment_1830310

F-4 Phantom pilot with a PhD in Climatology…a true Renaissance Man!

Jamie Hall

Even if a house isn't particularly draughty that wind will kill you.

Sal Santamaura

Interesting thread. I've nothing insightful to add to the many well-informed technical comments made. However,

@RascalOrnery "…We are way spoiled, generations before us grew up with outhouses in winter and a cook stove for heating. My father in law remembers ice on the inside of the windows…"

Perhaps it might be worthwhile to compare the life expectancy of those generations with current "spoiled" ones. 🙂

Hot_water_fan

Is spoiled the right word? It wasn’t available. How many people back then would choose to live that way if they could live like we do now? And vice versa?

Ironman

It’s already been alluded to in a couple of the responses, but design temperature almost always happens late at night, not during the day. The temperature is gonna rise during the day light hours and the mass of the structure will typically carry the temperature of the house through til then - even if the equipment is slightly undersized.

Folks also should understand that the heat won’t fail at design temperature: the structure will just fall behind maybe a few degrees until day time.

DCContrarian

Just for fun, here is the data for hours per year by temperature for Washington, DC:

I get this from the sizing tool at NEEP.org. There is a hole between 60F and 72F because that isn't either heating or cooling weather, which is all NEEP cares about.

Note that this line is not nearly the bell curve of a normal distribution, it seems to have two peaks and is not at all symmetrical. Plus there's that weird little "porch" in the low 20's. Note also that not only is the curve different for different locations, the shape of the curve is different in different places.

DCContrarian

OK, more fun with charts. This is a chart for efficiency vs. sizing for a particular heat pump, a Mitsubish PUZ-HA42NKA1 outdoor unit with a TPEADA0421AA indoor unit, performance data from:

https://ashp.neep.org/#!/product/156628/7/25000/95/7500/0///0

Efficiency, the Y-axis, is relative to the best number I could get from different configurations. The X-axis is sizing expressed as a percentage of the nameplate capacity, which is 54,000 BTU/hr.

I did an hour-by-hour analysis for a theoretical house in Suffolk County, New York. You can see the sheet here:

https://docs.google.com/spreadsheets/d/1d1Kox9yat2WqYWkqSIk-uqiJD0D-7eK3wQy1305sMCM/edit?usp=sharing

I varied the design load from 10,000 BTU/hr to 108,000 BTU/hr, oversized by a factor of 5.4 to undersized by a factor of 2. At each load I calculated the annual electricity usage, then divided by the design load to get a measure of efficiency. I then divided those numbers by the best ratio observed to score them as a percentage.

To my surprise, oversizing doesn't negatively impact efficiency. Undersizing does. The highest score is at about 20% of rated capacity, 20,000 BTU/hr. Below that efficiency drops, but barely perceptively. At rated capacity efficiency is about 97%, the line is basically flat below 40,000 BTU/hr, or about 80% of nameplate capacity.

This unit is rated for 38,400 BTU/hr at -13F, it seems that's about the capacity that should be used for sizing.

Jamie Hall

neat

DCContrarian

https://forum.heatinghelp.com/discussion/comment/1830393#Comment_1830393

Before we get too excited, I didn't capture the efficiency loss of a heat pump short-cycling, because I don't know how to calculate it.

LRCCBJ

https://forum.heatinghelp.com/discussion/comment/1830390#Comment_1830390

Well done.

My only complaint is the cost. For Suffolk County: $3643.86 @ $0.33/KWH. 1035g oil @ 80% @ $3.00= $3105.

After spending over $30K to buy it and install it, the oil fired boiler is less costly to operate at a fraction of the capital cost.

Why bother!

Kaos

@pecmsg I see this 100k load thrown around often. Around me that works out to about 2300 gallons of heating oil in a season. Are you actually going through that much?

Kaos

https://forum.heatinghelp.com/discussion/comment/1830412#Comment_1830412

The best I could find was for AC units. For those, there doesn't seem to be an efficiency hit for cycling (except for comfort and humidity issues). I would assume it would be similar for heat pumps

https://www.aceee.org/files/proceedings/2008/data/papers/1_9.pdf

Jamie Hall

https://forum.heatinghelp.com/discussion/comment/1830423#Comment_1830423

You might be surprised. Cedric goes through about 2500 gallons in a season.

DCContrarian

https://forum.heatinghelp.com/discussion/comment/1830424#Comment_1830424

Thanks. From the article: "This study found that in comparison to the standard practice
group, the resized group had longer average and maximum runtimes, consumed the same amount
of energy, and if implemented on a large scale would reduce the electric utility’s peak demand."

"Standard practice group" was oversized at 155% to 185% of Manual J, "Resized Group" was only oversized at 135% to 150% of Manual J.

If I read the article correctly, the claimed reduction in utility peak demand is because the smaller units can't actually keep up during the hottest hours, which is kind of not an apples-to-apples comparison.

The reason I ask about this is that NEEP.org, who I respect a lot, makes a big deal out of cycling. In their system capacity graph they identify hours where short-cycling is possible and include a calculation of "Percent Annual Load with Low-Load Cycling." They have a section of their website on "Oversizing Effects," but it just says oversizing is bad because it's oversizing, which is bad.

Jamie Hall

Personally I think that the downsides of cycling a system are both horribly magnified and over simplified.

What system? What's controlling the cycling? What is the cycling duty cycle? What is the overall frequency? And on and on.

But that's a topic for a whole different thread someday!

DCContrarian

Cycling is clearly more of a problem in cooling, because of the way dehumidification works. Air flows over a cold coil, and dew condenses out. When enough dew collects it starts to drip and is collected and disposed of. So at the beginning of a cooling cycle there is a delay before enough dew has collected to start dripping, and at the end of the cycle any moisture left on the coil will evaporate back into the air.

The longer the cycle, the more dehumidification takes place. Conceivably, cycles could be so short that no dehumidification happens, the sensible load has been met before enough dew accumulates to start dripping, and what's there just evaporates back into the indoor air. That would be most uncomfortable.

If you observe a heat pump, it's also clear that there is some inefficiency in cycling. It takes a couple of minutes of the compressor running to produce usable heat or cooling. But I have no idea how to quantify that.

DCContrarian

"If I read the article correctly, the claimed reduction in utility peak demand is because the smaller units can't actually keep up during the hottest hours, which is kind of not an apples-to-apples comparison."

I've been thinking about this. There is a fundamental asymmetry between heating and cooling with a heat pump. When you're heating, if you don't have enough capacity you can turn on resistance heat. There is no equivalent for resistance heat for cooling, if you don't have enough capacity you just have to let the house warm up.

So from a peak demand perspective, undersizing is good for cooling, everyone just let's their house warm up a bit. But for heating, you want people oversized, the last thing you want is everyone turning on their resistance heat at the same time on a cold day. And even at -13F cold climate heat pumps have a COP greater than 1.0, so the more heat you can get from heat pumps the better.

hot_rod

https://forum.heatinghelp.com/discussion/comment/1830438#Comment_1830438

would more cycles equate to more wear on the electro mechanical components?

Stop and go driving compared to interstate driving as an example.

All the components in a boiler, heat pump, AC, are engineered to a cycle life, motors, relays, gas valves.

When we build a zone valve the motor and end switch manufacturers ask us for an estimated number of cycles. They engineer and build to that spec.