Degree Days

Understanding Heating and Cooling Degree Days

Degree days are a specialist type of weather data, calculated from readings of outside air temperature. Heating degree days and cooling degree days are used extensively in calculations relating to building energy consumption, but the data is frequently used by those who don't understand what it really represents... This article aims to set that straight!

If you want degree-day data for a location near you, I suggest you head over to Degree  But if you want to understand what degree days really are (and personally I think you really should understand this before using them...), please read on.

There are three main types of degree days: heating degree days, cooling degree days, and growing degree days.  If you can understand exactly how one type of degree days work, it's very easy to understand the others.  So I've focused most of this article on explaining heating degree days:

Heating degree days

In a nutshell: heating degree days are a measure of how much (in degrees), and for how long (in days), the outside air temperature was below a certain level.  They are commonly used in calculations relating to the energy consumption required to heat buildings.

Why are heating degree days useful?

The energy consumption of building heating systems is more complicated than the energy consumption of TVs, kettles, or computers.  You can't just plug a heating system into a Kill-A-Watt meter to find out how much energy it uses each hour, because the energy usage of a heating system varies with the weather...

Essentially, the colder the outside air temperature, the more energy it takes to heat a building.

If you live in the Caribbean, it's probably warm enough that you won't need heating at all; if you live in New York, you'll probably only need heating in the winter; if you live at the North Pole... you'll probably want your heating on all year round.

But the outside temperature doesn't just vary from one location to another - it varies all the time, wherever you happen to be.  It's usually colder at night than it is in the day, and any single day/week/month/year is usually at least a little bit warmer or colder than the day/week/month/year before it.

If, like most people, you use your heating system to keep your building at a roughly constant temperature,  the amount of energy that your heating system uses will vary from one day/week/month/year to the next, just like the outside air temperature does.

Heating degree days are a simple way to quantify all of this.  The idea is that the amount of energy needed to heat a building in any day/week/month/year is directly proportional to the number of heating degree days in that day/week/month/year.

An example of heating degree days in action

(Skip this example if you have no desire to read my somewhat amateur attempt to bring the concept of heating degree days to life...)

Let me introduce you to a man called Dan.  Dan is the facilities manager of an office building, and he's under big pressure to reduce the building's energy consumption.  The company CEO, Jock, has noticed the rising cost of energy, and he's decided that the business could, and should, save some money by becoming more energy efficient.  Jock hasn't given much thought to how they're going to become more energy efficient, but he's certainly putting a lot of pressure on Dan to make it happen...

So, in January 2007, Dan spent a big chunk of his budget on improving the building's insulation.  At the time, he was confident that this would seriously reduce the energy it took to heat the building, and that the savings in the energy bill would very quickly pay for the rather hefty capital cost.

Roll forward to January 2008, a year after Dan's big insulation spending spree, and Dan has a decidedly stressed look on his face...  Jock, who's a "numbers guy", wants to see some "solid evidence" that Dan didn't "squander the company's hard-earned cash lining the pockets of some fly-by-night jokers".  (No offence intended to the insulation industry - Jock just tends to be a little quick to point the finger...)

Anyway, Dan is sweating, and it's not because the building's temperature control is playing up (the well-insulated building is actually helping to keep things at a steady, comfortable temperature).  Dan has just added up the heating energy consumption for 2007, and he's somewhat concerned by what he sees:

  • Heating energy consumption in 2006: 452,976 kWh
  • Heating energy consumption in 2007: 445,241 kWh

It's not that there hasn't been an improvement in the heating energy consumption (there has), it's just that Dan was rather hoping for more of an improvement...  After spending a small fortune on insulation, he was actually rather hoping for significantly more of an improvement...

Now, it just so happens that, in Dan's neck of the woods, 2007 was quite a lot colder than 2006.  Dan is aware of this, and, reluctant to admit that he might have overestimated the energy-saving power of his insulation idea, he is pinning his hopes on being able to prove that 2007's cold weather was to blame for the disappointing energy savings.  Dan tried explaining this theory of his to Jock, but he was met with a rather blunt "Don't you try fobbing me off with any of your hand-waving nonsense!"

Shame on you, Dan, for forgetting that Jock is a numbers guy...

Fortunately all is not lost, as a colleague has tipped Dan off to these things called heating degree days.  They're basically a measure of how cold the temperature was, but they're specifically for heating - if you've got 10% more degree days in any day/week/month/year you should expect 10% more heating energy consumption in that day/week/month/year, all other things being equal.

So, Dan hunted the web until he found Degree, a site that generates degree days for locations around the world.  He found a weather station near the office building, and downloaded a few years' worth of heating degree days for that location.  He quickly assembled the following figures:

  • Heating degree days in 2006: 3,320 (I'll explain what this number really means shortly)
  • Heating degree days in 2007: 4,092

Applying some simple arithmetic:

  • kWh per degree day in 2006 = 452,976 / 3,320 = 136
  • kWh per degree day in 2007 = 445,241 / 4,083 = 109

Comparing these two figures, Dan concluded that the heating energy efficiency in 2007 was around 20% better than that in 2006.  Well done Dan: your insulation plan was a good one, and the company should make good savings from it for many years to come.  In fact, it was such a good idea, Jock has convinced himself that it was his idea all along...  So, Dan, it's looking unlikely that your insulation success will help your bonus, but at least you can stop sweating - your job security is no longer in immediate danger.

(Advanced note: the kWh-per-degree-day calculations above are a very simple way to estimate energy savings. I don't want to overload you in an introductory article, but, just so you know, it's usually better to use a more sophisticated regression-based approach to calculating energy savings. I suggest you keep reading this introductory article for now, but remember to read about that better approach when you are ready to do similar calculations of your own.)

So how are heating degree days calculated?

Well, there's one correct way to calculate heating degree days (which requires vast quantities of temperature data - infinite quantities to be precise), and numerous different ways to approximate the same result using less temperature data.

Nonetheless, irrespective of the exact calculation method, it always starts with a base temperature:

The base temperature of a building

With regard to heating degree days, the base temperature of a building is the temperature below which that building needs heating.
Let's consider a regular office building.  In fact, seeing how I put all that effort into the Dan story, let's consider the office building that Dan the facilities manager is in control of.

Dan tries to keep the office building heated to around 20C (about 68F) - after many years on the job he has determined that this is the temperature at which he gets the least number of people complaining that it's too hot or too cold.

On a summer day, when the outside temperature is 20C or above (about 68F), as you can probably guess, Dan switches the heating off - there's no point in heating a building when it's already warmer than the temperature you want it.

In fact, Dan has figured out that he can switch the heating off when the outside temperature reaches 17C (62.6F) - a few degrees below the desired inside temperature.  The office has a lot of warm people in it and a lot of warm office equipment too - this essentially provides a few degrees of free heating.  In technical terms, this would be described as an average internal heat gain of 3C, or 5.4F.

So, when the outside temperature is below 17C (62.6F), the heating needs to be on, and when the outside temperature is above 17C (62.6F), Dan can switch the heating off without incurring any more complaints than usual about it being too cold.  Of course more people might complain that it's too hot, but that's a different story.

What this means is that the base temperature of Dan's building is 17C, or 62.6F.

All buildings have a base temperature - it varies from building to building, but you can think of it as depending on two things:

  1. What temperature is the building heated to?  (e.g. Dan's building is heated to 20C or 68F.)
  2. How much free heating comes from the people and equipment inside the building?  In other words, what's the average internal heat gain?

The base temperature of your building will determine the base temperature of the heating degree days that you should use to do your calculations.

(Advanced note: In reality it's more complicated than that - the best base temperature to use can also depend on the thermal properties of the building, the heating schedule, and external influences like solar gains.  Considering the two factors above can help you get a reasonable first estimate, but at some point you should read this article on choosing base temperatures for more detail.  If you've got good records of the building's energy consumption, it's also a good idea to estimate the base temperature experimentally using linear regression analysis, and Degree offers a regression feature that can do this for you automatically.)

Anyway, that explains the base temperature...  But what do the heating-degree-day numbers actually mean?  To understand this, you need to have a rough idea of how the figures are calculated.

Turning temperature readings into heating degree days

With the appropriate use of big, scary-looking formulae, it's quite possible to make it look as though degree-day-data calculation is something that's best left to the experts. But it's actually very straightforward to turn temperature readings into degree days.  I'm going to use a few example calculations to explain how the process works for heating degree days.

Let's say that we're dealing with a building with a base temperature of around 17C (I'm going to stick with Celsius for this explanation, otherwise it'll get really confusing).  It's the start of July - technically it's the middle of summer, but the inhabitants of the building are still waiting for "the real summer" to arrive.  (I'm from the UK, where it's depressingly common to spend most of the summer months waiting for "the real summer" to arrive...)

Anyway, consider a single day, let's say July 1st, when the outside air temperature was 16C throughout the entire day.  A constant temperature throughout an entire day is rather unlikely, I know, but degree days would be a lot easier to understand if the outside air temperature stayed the same...  So, throughout the entire day on July 1st, the outside air temperature (16C) was consistently 1 degree below the base temperature of the building (17C), and we can work out the heating degree days on that day like so:

1 degree * 1 day = 1 heating degree day on July 1st

If, on July 2nd, the outside temperature was 2 degrees below the base temperature, we'd have:

2 degrees * 1 day = 2 heating degree days on July 2nd

Let's look at July 3rd - this was a hotter day, and the outside air temperature was 17C, the same as the base temperature (i.e. 0 degrees below the base temperature).  This gives:

0 degrees * 1 day = 0 heating degree days on July 3nd

On July 4th it was warmer again: 19C.  Again, the number of degrees below the base temperature was zero, giving:

0 degrees * 1 day = 0 heating degree days on July 4th

You might have guessed: when the outside air temperature goes over the base temperature, you don't get any heating degree days.  This makes sense, because you wouldn't need any heating either.

Right, now let's make it a little more realistic.  July 5th had a temperature of 15C from 00:00 to 12:00, and 16C from 12:00 to 24:00.  So for that day we have:

(2 degrees * 0.5 days) + (1 degree * 0.5 days) = 1.5 heating degree days on July 5th

(The 2 degrees is because 15C is 2 degrees below the base temperature of 17C, and the 0.5 days are because 00:00 to 12:00 is half a day.  We calculate the heating degree days for each period in the day, and then add them together to get the total for that day: 1.5.)

On July 6th, colder weather started moving in: the temperature was 16C from 00:00 to 06:00, 15C from 06:00 to 12:00, 14C from 12:00 to 18:00, and 13C from 18:00 to 24:00.  This gives the following:

(1 degree * 0.25 days) + (2 degrees * 0.25 days) + (3 degrees * 0.25 days) + (4 degrees * 0.25 days)
= 2.5 heating degree days on July 6th

Now, on July 7th, the temperature just kept changing... like it might on a real day...  Between 00:00 and 00:30 it was 13C, between 00:30 and 01:00 it was 12.9C, between 01:00 and 01:30 it was 12.9C, between 01:30 and 02:00 it was 12.8C... it started getting warmer around 05:00, peaking at 17C between 14:00 and 14:30, and dropping again until it reached about 13.7C between 23:30 and 24:00.  Complicated!

A proper calculation would not make for particularly interesting reading, so I'll leave most of it out.  But essentially you just have to add up the figures for each of the half-hour periods in the day (one half-hour period is 1/48 days):

(3 degrees * 1/48 days) + (3.1 degrees * 1/48 days) + ....... etc.
= 1.9 heating degree days on July 7th

Hopefully by now you're getting the idea!

So, from the examples above we've got:

  • July 1st: 1 heating degree day
  • July 2nd: 2 heating degree days
  • July 3rd: 0 heating degree days
  • July 4th: 0 heating degree days
  • July 5th: 1.5 heating degree days
  • July 6th: 2.5 heating degree days
  • July 7th: 1.9 heating degree days

We'd expect the heating energy consumption on each of those days to vary with the heating degree days.  So, the heating on July 2nd would use twice as much energy as the heating on July 1st, and, on July 3rd and July 4th, the heating wouldn't use any energy at all (zero degree days on those days would mean it would be warm enough for the heating to be switched off).

One of the best things about degree days is that you can add them together.  Adding together the readings above gives a total of 8.9 heating degree days for the week beginning on July 1st and ending on July 7th.  So we'd expect that the heating system would have used 8.9 times more energy in that whole week than it used on July 1st alone.

If you've got daily heating-degree-day values for each day in a month, you can add them up to get the total heating degree days for that month.  And if you've got the heating-degree-day values for each month in a year, you can add them up to get the total heating degree days in the whole year.

And therein lies what I consider to be the beauty of degree days: you can add them up to get totals for long periods of time, and they still represent all the relevant variations in temperature over that whole time period.  (Contrast that with an annual average temperature, which would tell you nothing about how much the temperature varied within that year.)

Real-world calculation methods

The calculation method that I explained above is essentially the correct one for calculating heating degree days: for each period over which the outside air temperature was constant, you multiply the degrees below the base temperature by the number of days that the temperature was fixed for (usually small fractions of days), and then you sum all the values together to get the total heating degree days for the period in question.

The problem with that approach is that, in the real world, outside air temperature doesn't remain constant - in fact it changes pretty much all the time.  Mathematically speaking you'd need an infinite number of temperature readings to calculate degree days properly.

Fortunately, "mathematically speaking" doesn't really matter too much in this instance, and half-hourly or hourly temperature readings are plenty good enough to calculate degree days accurately using the method described above.

However, reliable half-hourly and hourly temperature readings are rarely readily available, so there are a number of other approximation methods that are used to calculate degree days from more commonly available measurements of outside air temperature.  These methods typically use either the daily maximum and minimum temperatures, or the daily average temperatures.

Personally I'm of the opinion that the the details of the approximation method used are not important, so long as it uses the data it's given to generate degree-day figures that are very close to those that would be generated by the correct method (or, more realistically, by a method that used half-hourly temperature readings or similar).

A note on common base-temperature confusion

Unfortunately, when it comes to degree days, confusion is abundant, even amongst those that really should know better...  So a lot of people are unaware of the fact that degree days can come in any base temperature, and that typically you should choose the base temperature to fit the building that's energy consumption you're looking at.

Earlier I gave the example of a particular office building with a base temperature of 17C (62.6F).  This is probably a fairly common base temperature for an office building, although the actual base temperature of any particular office building will depend on what temperature it's heated to, and what people and equipment are in it.

If your building happens to be a swimming pool hall, the base temperature might be more like 26C (78.8F), and you'd want to use a very different set of heating degree days for your calculations.  Or maybe you have a just-warm-enough-to-prevent-a-lawsuit factory, or a house that you like to keep uncommonly warm - whatever the building, it will have its own base temperature, and you should use heating degree days with that base temperature when you're looking at its energy consumption.

So it's a popular misconception that degree days come with a single base temperature.  In the US, this is typically quoted as being 65F, whilst in the UK it's typically quoted as being 15.5C.  I suspect this stems from the days when degree days were made available in magazines and the like, and there was no more room in magazines to fit data with a range of base temperatures than there was the computer technology to calculate and store all that data in the first place.

So the next time you hear something along the lines of "heating degree days have a base temperature of 65F", you will hopefully know better than to believe it!

(All that said, it makes perfect sense to use a consistent base temperature when you're comparing the climate of one location with that of another.  But when you're looking at the energy consumption of a building, there's no doubt about it: you should use degree days with the most appropriate base temperature for that building.)

Other types of degree days

Hopefully by now you have a good understanding of what heating degree days are, and how they are calculated.  Now it should be very easy to understand the other two types of degree days - cooling degree days and growing degree days:

Cooling degree days

Cooling degree days are a measure of how much (in degrees), and for how long (in days), the outside air temperature was above a certain level.  They are commonly used in calculations relating to the energy consumption required to cool buildings.

I think of them as heating degree days in reverse: whilst heating degree days start adding up when the outside air temperature drops below the base temperature, cooling degree days start adding up when the outside air temperature rises above the base temperature.  So the base temperature of cooling degree days is just the temperature above which the building needs cooling.  Pretty straightforward, right?

Growing degree days

In their simplest form, growing degree days are calculated in the same way as cooling degree days, but the base temperatures used are based upon the temperatures above which certain plant or insect growth occurs.

Different plants have a different base temperature above which they will start to grow, and their growth will typically be roughly proportional to the amount by which that base temperature is exceeded.  This is very similar to the way in which building cooling is proportional to the amount by which the building base temperature is exceeded.

Unlike heating and cooling degree days, growing degree days sometimes have cut-off points (e.g. when the temperature gets over 65F we stop counting - that sort of thing).  To be perfectly honest, my expertise are in the heating and cooling side of degree days, so I won't try to elaborate much further on growing degree days!

Final degree-day advice

Although it's important to understand how degree days are calculated, there is rarely a need to actually calculate them manually.  My company runs a website called Degree that generates heating and cooling degree days for locations worldwide, to any base temperature, and there are a number of other sources of degree-day data too (albeit sources that focus on specific countries or regions).

One final point: however accurate the degree days are themselves, it's important to understand that the calculations that use them are usually only approximate - there are a number of problems with the commonly-used applications of degree days that I've not discussed here at all.  I wrote a detailed article about these issues called Degree Days - Handle with Care! (yes, that exclamation mark is important!) - you might be interested to take a look at it now you have a good understanding of the basics.

Anyway, best of luck with all your degree-day based work, and I hope that it helps you to save energy!
Martin Bromley


Predicting Heating Degree Days

Great writing, Martin.
Your examples refer to using HDD for looking at historical figures - can they be used to forecast heating degree days and therefore energy costs, based on the weather forecast (along with my home's base temp and my desired temperature)?

Last edited Oct 17, 2011 4:33 AM

Thanks Sylvia!

HDD can be derived from forecast temperature data in the same way that they can be derived from historical temperature data. So, yes, in that sense, they can be used to forecast heating energy consumption.

However, I'm not sure how feasible it is to forecast temperatures with a level of accuracy that would be useful for analysis over the sorts of timescales you might want for predicting heating/cooling energy consumption (like the next few weeks or months as opposed to just the next few days).

People often use averaged historical degree days, like 5-year or 10-year average data, as a predictor of future consumption. So, with now being October, you could look at the 5-year average for November to get an approximate idea of what to expect next month. Of course that can only give you a rough idea, since degree days in any given month can vary quite a lot from year to year (more in some climates than others). But looking at averaged degree days is a better predictor than just looking at last year's degree days. You can also do things like consider the coldest November (the one with the highest HDD) in the last 10 years as a likely worst-case scenario, and the warmest November (the one with the lowest HDD) as a likely best-case scenario.

Posted by Martin Bromley, Oct 17, 2011 4:33 AM

strong intra-day temperature variation

thanks for this article. If I'm correct, long ago degree days were not computed based on integration of the temperature difference of the base temperature and the outdoor temperature but based on the average of daily mininum and maximum temperature. I guess this was before the existence of affordable computers... Of course, this can give completely different result.

Also at that time, and following from the calculation method, degree days were only taken into account if the average daily outdoor temperature was below the day temperautre. And that's now my question: is this still done like that today or not?

So, to give a practical example: suppose a base temp of 15°C, and a daily temperature of 14°C for 8 hours in the morning, the rest of the day is above 15°C and the average is also above 15°C. In a building with thermal inertia, you might not need to heat that day (also depending on solar radiation). Does this day have 0.33 degree days in today's calculation method or not?

Thanks a lot for your answer,

Last edited Mar 28, 2011 9:32 AM

Roel, that's a good question.

For your example, the proper calculation method would indeed give you a result of 0.33.

Some sources do still approximate degree days by taking the average of the daily max and min and taking the difference between that and the base temperature. We call that the Mean Temperature Method. Its easier to calculate data that way than to do it properly, and the ease of calculation is particularly compelling for sources that don't specialize in degree days or energy consumption. It's not such a bad approximation when the temperatures don't cross the base temperature in a single day (in fact it can be very good if it uses a real average rather than max - min / 2), but it's lousy when they do.

Ideally degree days do correlate with energy consumption (HDD with heating energy consumption and CDD with cooling energy consumption), such that, if the degree days are zero, then the heating/cooling energy consumption would be zero too. Though, as explained in the article at things get messier when the temperature is near to the base temperature and the degree days are near to zero. For starters, the choice of base temperature makes proportionally more of a difference to the calculated figures (irrespective of calculation method), and the base temperature is only an approximate concept in the first place.

With the particular example you give, assuming the building only needs heating after, say, 9am, you could argue that the Mean Temperature Method would work well for the hypothetical temperature variation you described. But really that would just be a happy coincidence for that particular temperature/heating configuration. A lot of buildings are heated only during the daytime, but a lot of other buildings are heated in the evening/night as well or instead, or 24 hours a day. And for cooling, that happy coincidence could only arise for buildings that are cooled in the night-time only. When higher temperatures in the day coincide with a daytime-only cooling pattern, the Mean Temperature Method is particularly hopeless, since the common case of a cold night that's balanced by a hot day (requiring some considerable cooling) gives degree days of zero.

Intermittent heating or cooling (intermittent meaning not 24/7) makes modelling the HVAC energy consumption of buildings considerably more complicated. The effects of thermal inertia (which effectively introduce time lags between the outside temperature and the inside temperature) don't help either.

Aside from much more sophisticated modelling (including modelling various building properties like thermal admittance and air exchanges), the practical solution is typically to adjust the base temperature. Intermittent heating has the overall effect of lowering the base temperature for heating degree days. For example, take two buildings with identical construction... Building A is occupied 24/7 and Building B is occupied 0900-1700, Monday to Friday only. Assuming both are heated to the same temperature when occupied, then Building B will have a lower base temperature overall. In other words if, through regression analysis, you find the optimal base temperature of both buildings, you should find that the base temperature of Building B is lower than the base temperature of Building A.

This is all quite hard to explain, so my apologies if it's not very clear!

Posted by Martin Bromley, last edited Mar 28, 2011 10:38 AM

Building Base Temperature and R2

Excellent Article, thanks.

One question if I may. I read in the blurb in the DegreeDays site that if you do a mathematical spread of R2 numbers across all the range of base temperatures, the column that showed nearest to 1 would correlate to the optimum base temperature of my building. If this is the case, then the optimum figure for me would be 12.5, which I would have thought was low for a hotel. Have I understood this correctly? (the kWh figures I used were for heating consumption only, not taking into account the Hot water generation or cooking gas)

Last edited Jan 5, 2011 3:52 AM

I'm a fan of the R2 approach described at - it's good to be able to base a base-temperature decision on data rather than solely intuition. But that method isn't foolproof so it's important to make sure the base temperature you pick makes intuitive sense as well.

A base temperature of 12.5C is not unheard of for a well-insulated building, particularly one with intermittent heating where the heating is switched on and off to fit building occupancy (this has the effect of lowering the overall base temperature). For a hotel that's kept at a relatively high temperature throughout, for 24 hours a day, 7 days a week (i.e. not intermittent heating), I would be surprised if the real base temperature was 12.5C (unless perhaps the hotel was exceptionally well insulated). But it's impossible to generalize too much about hotels since the buildings and heating/usage patters can vary a lot. If your hotel was, in part or in whole, closed and largely unheated for a couple of days a week, that could lower the base temperature quite a bit.

Posted by Martin Bromley, last edited Jan 5, 2011 3:52 AM

Love the way you explain


Last edited Feb 14, 2012 5:28 AM


First, Thanks! Now my company in Chicago became a DOE Save Energy Now LEADER and i am new to this Degree Days.

I have done much research but still cannot understand a proper HDD base temp because of the smart thermostats that can be set during different times of day. For instance, during the day we keep the building heated at say 70 in the office (65 in warehouse) but at night we put it at 55 degrees from 17:00-5:30. As we all know the outside temperature is colder at night than during the day. That is more than half the day at 55 degrees! So what i did, not sure if i am right, but i picked 62 as a number in between the heat of people and the colder night temps? The heat does not go on in early fall or late spring just because of the night being under my base number. It is near 62 degrees during the day, we will have to put on the heat.

Am i on the right track?

Last edited Sep 16, 2010 3:42 PM

My favourite approach to picking a base temperature is an experimental one: see for more on that.

Based on what you've said about your building, your estimate of 62 sounds like a reasonable estimate to me, so, yes, I think you're on the right track. But you might find that you can improve the correlation a little if you give the experimental approach a go and try correlating your energy consumption with HDD with base temperatures around that point. The tighter the HVAC control in your building, the more likely it is that you'll get useful results through the experimental method.

You might also find a couple of the FAQ at useful.

Good luck!

Posted by Martin Bromley, last edited Sep 16, 2010 3:42 PM

Very good article

Thank you for a very informative and easy to read article. I'm not sure about your assertion that the beginning of July is the middle of summer in the UK though, I think it's officially somewhere around the beginning of August.

Last edited Jan 6, 2010 8:02 AM


Re: when the middle of summer is... I just spent some time getting confused by Wikipedia's explanation of the dates of seasons. It seems there are several definitions of "summer". Going with the definition that says summer starts at the beginning of June and ends at the end of August, mid July would be the middle of summer. Though to be honest I'm not sure which definition is the one to use.

The longest day of the year, sometimes called midsummer day, is June 22 or thereabouts, so that would be a reason to define the middle of summer as being earlier than I put it rather than later...

Posted by Martin Bromley, last edited Jan 6, 2010 8:02 AM

Using DD to Calculate Heating Cost of Exhausted CFM

How would you go about calculating the annual energy savings during the heating season only of reducing the operation of bathroom exhaust fans from 24 hrs a day and 7 days per week to 8 hours a day and 5 days a week? The basic equation is BTU/YR = CFM x DD (heating only) x 24 x 1.08 if the fans were to operate 24 x 7.

Last edited May 1, 2009 2:08 AM

That's a tricky one...

Calculating the reduction in electrical consumption of the fan itself should be relatively easy, but the problem is that, when the building is heated, the fans are expelling warm inside air and replacing it with cold outside air. I'm no expert on bathroom fans, but I understand they're often sized to increase the air changes per hour to something like 8. Which is a lot higher than, say, 0.5 to 2 (ish), which is a fairly common ventilation rate (air changes per hour) in many buildings.

The heating energy consumption of a building is essentially due to
a) heat loss through the building fabric (walls, roof, floor); and
b) heat loss through the air changes

The proportion of each depends considerably on the construction and operation of the building, but, if I recall correctly, I think it's pretty common for b) to comprise a big proportion (like 50% or more).

The energy consumption of intermittently heating buildings is complicated, as a lot depends on how the building retains its heat when the heating is switched off.

So you're talking about reducing the operation of the exhaust fans from 24/7 to typical office hours. The reduction in heating consumption that would result from that would depend greatly on whether the building is heated 24/7 or not. If it's not, then the heating savings would probably be pretty small (depending on how much the exhaust fans accelerated the temperature drop of the building overnight). If it is, then I think you could very approximately assume that the consumption would be reduced proportionally to the reduction in hours of usage (minus a bit for the fact that temperatures in the daytime hours of operation are higher than night-time temperatures when the fans are now switched off - see the handle with care article section on intermittent heating for more on this issue).

Though that's just my best estimate - you might do well to go back to the source of the equation you quoted to see if they have any rules of thumb for intermittent heating. Intermittent heating is complicated enough that people often sidestep the "understand the building" stage and use a rule of thumb (often based on experimental results rather than theoretical) to estimate things like this.

Posted by Martin Bromley, last edited May 1, 2009 2:12 AM

Very nicely explained

Thanks Martin, for your patience with my novice questions. This is well explained and hopefully it's now gone into the thick spot upstairs.

Last edited Apr 10, 2009 12:53 PM

Well Done!


Thank you for all of your hard work and dedication. I feel as if I have definitely absorbed yet another knol.

Mar 4, 2009 6:32 PM

Excellent, but have a question

This was extremly informative, but would like to know if I can apply this (HDD base temperature) to an outside heating source. Trying to calcualte efficiency of a heated walkway, melts snow in a village of a ski resort. The system is always on unless the outside temp hits 42 F, in which case it turns off. Could I use 42f as the base temp to get my HDD for this 'building'?. Thanks for any help on this. On to read HDD - Handle with Care..

Last edited Feb 9, 2009 3:51 PM

Hmm, that's an interesting question... I think the applicability of HDD for the walkway would depend on how clever the walkway was. If it's not very clever, then it'll use the same amount of energy when the outside temperature is 40F as it would when the outside temperature is 10F, because at either temperatures it would be switched on (heating) constantly. If that's the case, you couldn't really treat it as a building.

But is it possible that the thermostat that controls the walkway is positioned on or right next to the walkway itself (i.e. measuring the temperature of the heated walkway rather than the outside air temperature)? If that's the case, the outside temperature could be -10F, and the walkway would still switch itself off for a while or reduce its heating consumption if it managed to heat itself up to 42F. Then I think HDD might correlate pretty well with the energy consumption of the walkway. And, although I know nothing about heated walkways I'm guessing that might be the way one would operate...

If you have measurements of the energy consumption of the walkway then the easiest thing to do might be to try correlating those measurements with HDD to see if there's a pattern. In technical terms what you'd need would be a regression analysis - basically a plot of kWh (or any other measurement of energy consumption) against degree days. The Handle With Care article has a bit more on this.

Posted by Martin Bromley, last edited Feb 3, 2009 12:52 PM
Hi Martin, thanks for your prompt reply! This heated walkway is pretty clever, it heats the slab (walkway of 30,000 sq ft) to 28F then idles until a built in sensor in the ground gets covered in snow in which case it kicks on the snowmelt system and heats the slab until it melts what is covering the sensor, or it will continue to heat if the slab temp drops below 28F, once it reaches 28F again it idles. If the outside air temp reaches 42F the warm weather cut off kicks in and it switches off. We use propane to heat the water and glycol mixture that flows through the tubes under the walkway and we are trying to normalize the meter readings from month to month and year over year and take into consideration the number of days between meter reads each month vs the same month in the prior year, and the biggest variable is the weather. Regression sounds like a good plan, and will help us predict our consumption. One last question, are you suggesting that I NOT treat this as a building and use 42F as the base temp, but just do a regression?
Thanks for all your thoughts and wisdom! Susie

Posted by SThiele, last edited Feb 6, 2009 10:52 AM
Hi Susie, from what you've said I think one could argue that a base temperature of 28F would be appropriate. Comparing the slab to a building, 28F is the temperature above which heating isn't needed, so it'd be similar to the base temperature of a building in that sense (I'm guessing the concept of average internal heat gain wouldn't be applicable to the slab...).

If it wasn't for the fact that the heating system idles between 28F and 42F, I think it would make perfect sense to use HDD with a base temperature of 28F. The idling complicates matters as the energy consumption won't be temperature dependent within that range (it will be constant). If it wasn't for the 42F cut-off, the idling energy consumption would be the equivalent of a building's baseload energy consumption (and so would be handled by the regression analysis), but because the system cuts off at 42F, that might mean the regression analysis with 28F HDD won't work so well.

But hopefully the idling energy consumption is very small compared to the consumption when the temperature drops below 28F (and the slab starts heating properly), so hopefully it isn't something to worry about. But you might find that HDD with a base temperature a little over 28F give a better correlation than 28F HDD - adding on a few degrees might account for the idling to some extent, in a very approximate and non-scientific way!

I would suggest that you give the regression analysis a go with some past records of consumption, and see if you get a reasonable correlation between propane consumption and HDD. Bearing in mind that your regression analysis will require HDD, and those HDD will be of a particular base temperature (all HDD have a base temperature). Try it with HDD with a base temperature of 28F, and see how good the correlation is. If it's going to work you should see a pattern just by looking, but, if you're using Excel, you can also add a trend line to the scatter chart and add an R2 value, which will give you an indication of how good the correlation is.

You might find that you get a better correlation with HDD with a base temperature somewhere between 28F and 42F, so you could try correlating with a range of base temperatures to see which works best.

It'll be approximate (like pretty much any calculaion involving degree days), but if you get a reasonable correlation you should be able to use that to roughly predict your consumption.

Fingers crossed it works well... Good luck!

Posted by Martin Bromley, last edited Feb 9, 2009 3:51 PM

Very helpful introductory resource

Thanks so much! However, there is at least one more type of degree-day: infiltration degree days (IDD). I'd appreciate if you would also include information about IDD, as it is pertinent to seasonal energy calculations, according to the Max Sherman 1995 paper "Uses of Blower-Door Data."

Last edited Jan 20, 2009 4:56 AM

Hi Jane, thanks for your comment and the tip on IDD.

I'm ashamed to admit that IDD are new to me - I shall have to look into them further. Please say if you know of any good references that explain exactly what they are.

One thing I can say about IDD is that they don't seem to be particularly widely used. As a quick, very approximate test, I looked to see how many mentions Google had for the various types of degree day:

"heating degree days": 1,090,000
"cooling degree days": 523,000
"growing degree days": 160,000
"infiltration degree days": 147

Based on those numbers, I'm not sure that an explanation of IDD would be suitable for inclusion in this introductory article. But if there's a good page elsewhere that explains them clearly I'd be happy to add a link to it.

Posted by Martin Bromley, last edited Jan 12, 2009 9:01 AM
Hi Martin,

IDD were developed in the Lawrence Berkeley Laboratory. The major document that describes them is Max Sherman's 1995 paper on Uses for Blower-Door Data: The explanation is very dense, however.

Posted by Jane Yang, last edited Jan 12, 2009 10:08 PM
Thanks Jane. And sorry for taking so long to say thanks... Google sends me an email when somebody comments on this knol, but it seems they don't when somebody writes a comment under a comment.

Posted by Martin Bromley, last edited Jan 20, 2009 4:56 AM

Enabled me to get a baseline reading

Thanks for the article, and the degreedays site. It's enabled me to get a nice baseline reading for our new house. I've linked to both in my blog posting:


Last edited Jan 6, 2009 8:08 AM

Thanks for the mention on your blog M@!

Glad the baseline is working out. As you gather more readings I'd suggest experimenting with HDD with different base temperatures to see if they help you improve the correlation. Your choice of base temperature will probably make a surprising difference to the gallons-per-HDD figures you get for different seasons.

Posted by Martin Bromley, last edited Jan 6, 2009 8:11 AM

Simply explained and informative

I love well-done explanations of any subject and amusing examples always help! I only came here for the data but had to give my compliments on both the application (very easy to use) as well as the articles you've written. Cheers!

Last edited Dec 22, 2008 8:16 AM

Much appreciated - thanks Karl!

Posted by Martin Bromley, last edited Dec 22, 2008 8:16 AM

Pretty well explained...

Well done in taking time and using rigourous terms to define.

I would just say that you forgot to tell explicitly that the setting point to consider should be equal to the base temperature plus the equivalent raise of temperature given by thermal gain for heating for the building, with an off-setting of a few hours depending on building inertia.

Let's hope people and especially company chief executives will reward facilities manager for proving their proficiency to reduce Green House Effect through energy consumption, reduce equipment size dimensioning providing space gain, more comfort thanks to insulation so money and environment friendship is at that (small) cost.

Last edited Dec 21, 2009 4:24 AM

Thanks for the comment!

"I would just say that you forgot to tell that the base temperature should be the setting point for heating for the building, with an off-setting of a few hours depending on building inertia."

If you have a moment, could you explain this point in a little more detail?

Posted by Martin Bromley, last edited Dec 10, 2008 5:40 AM
HI, sorry for replying so late.

Here are the explanations of my ideas because people would think base temperature is the only temperature you use to decide when heating a building :

1. Your base temperature (based on outside temperature) is different from the setting heating temperature (based on a measure indoor). You use both to regulate inside temperature.
2. Building has inertia in energy transmission (I am sure you know that) so Imagine you have a few hours in the daytime under your base temperature but above after for same time. Result : people won't feel (probably except if warehouse-like building with very light structure) a difference in inside temperature evolution although you measured some degree days !

So although this idea is going into more detailed reflexion, I think that UK buildings must not regulate with proper formulae because energy consumption is high compare to your mild weather (I am from France) : about 400kWh/m².year against 280kWh/m².year for offices.

Hope was interesting, because I don't have your writing talents.

Posted by PRIN Jonathan, last edited Apr 26, 2009 8:45 AM
Sorry for taking such an incredibly long time to respond - Google Knols doesn't send me an email notification for replies to comments so they're easy to miss.

Anyway, thanks for the explanation. It's a good point. Building heating and cooling is seriously complicated... I've actually spent a lot of time developing software that attempts to model it - still a work in progress! Modelling heating and cooling of a building would be a whole lot simpler if it wasn't for the inertia that you describe.

I think it's difficult to account for all the relevant thermal properties of a building when estimating the most appropriate base temperature. Usually I think it makes most sense to correlate real consumption figures with degree days from multiple base temperatures to see which gives the best correlation (and is therefore likely to be the most appropriate to use in future). I'll see if there's a good place that I could edit the article to say something along these lines without confusing people too much!

Posted by Martin Bromley, last edited Dec 21, 2009 4:24 AM

Well done!!

I actually found this knol after I read the "Handle with Care!" article, and found both to be very rewarding. (Then again, I'm the grown daughter of an energy professional, so I'm quite familiar with the terminology.)

This article, however, was great for the way it made the information accessible and friendly -- I liked how Dan had to search out a way to find some real numbers to report to Jock. My similarity to Jock is precisely why I am here today: I'm looking up heating degree days to put into my home energy analysis spreadsheet! :)

So, thanks for this, and for

Last edited Nov 26, 2008 12:15 PM

Thanks for your kind words Crystal, and thanks also to everyone else that's commented.

I'm particularly glad to hear when people have found the "Handle with Care!" article useful, not least because it took a long long time to write! I think the concepts it covers are quite a bit more complicated than the ones that this knol covers, and I found it quite hard work to explain them all well. Although I think the information that's in that article is well worth understanding, I think it would be a better article if it was written in a way that was less heavy going, and more enjoyable to read... Right now I suspect that, despite it being useful to many of the people that do go though it all, it's also fairly good at dispensing headaches and inducing sleep... I'm guessing it might have that effect on me if I hadn't written it :)

Posted by Martin Bromley, last edited Nov 28, 2008 2:40 AM


Pretty understandable, moving on now to Handle with Care!

Last edited Nov 5, 2008 12:09 PM

Much Appreciated

Thank you for your very clear, and fun, explanation. Are you in the southern hemisphere? Up here in New York City, it would be unlikely to have HDD in July.

Last edited Sep 24, 2008 4:08 AM

Thanks for your comment Carmen (and thanks to everyone else that's commented too). I'm based in the UK where, much to my dismay, we get tend to have at least some HDD in every month of the year (with a base temperature of 17C we do anyway). I have family in New York state and I've spent quite a lot of time there - I envy your summers!

Anyway, I've made a small edit to the article to explain why my July temperatures are so cold.

Posted by Martin Bromley, last edited Sep 24, 2008 5:17 AM

Very Helpful


Last edited Sep 7, 2008 6:17 PM


Great document...


Last edited Aug 30, 2008 8:32 AM

An interesting concept

I can see where this would fit in nicely with the Building Controls and Automation Industry to increase the efficiency and allow Facility Managers to plan ahead. Good show!

Aug 22, 2008 1:16 PM

Nicely explained

You've done a nice job in explaining something complicated in a clear and simple way. Great work!

Last edited Aug 18, 2008 7:07 AM