HDDs will give you a good first estimate, but as we have discussed before, they are not brilliant.

The usual method is to plot the kWh against HDDs (kWh on the y-axis, HDDs on the x-axis).
Then add a Trend line and extend it backwards until it crosses the y-axis (easy enough in Excel).
The point where is crosses the y-axis is your fixed loads, in your case DHW. This will give you your daily kWh usage.
You can also just separate out the non space heating period and average that (total kWh used divided by days). That will give you your kWh/day.
As you know the energy delivery rate (3.1 litre/hour or 31 kWh.hour) you should be able to estimate your burn time.

So if you use 8 litres of oil a day. 8 litre / 3.1 litre per hour = 2.58 hours.

How about an 'hours run' meter on the mains supply to the burner oil pump?

Anyone have any thoughts on an appropriate base temperature for well-insulated buildings?

Carbon Trust uses 15.5C on the assumption that when external temperature is above 15.5C most buildings need no heating. http://www.carbontrust.com/resources/guides/energy-efficiency/degree-days

However, it can get a lot colder than that, before a well-insulated building needs any heat additional to that provided by bodies, pets, lights and appliances.

I am not arguing for a lower base temperature. On the contrary, it might be better to use a base temperature of, say, 20C, and explicitly subtract the 'other' sources from the calculated heat demand.

I'm sure you Passivhaus peeps are across this one

Posted By: rhamduI am not arguing for a lower base temperature. On the contrary, it might be better to use a base temperature of, say, 20C, and explicitly subtract the 'other' sources from the calculated heat demand.

I'd agree with that - particularly if you can make some estimate of the solar gain appropriate to the times of year in question.

But - you then really need to do things at higher time resolution than per month down to of the order of the heating timeconstant of your building. A month with the temperature at 15.5 °C where you just about don't need heating is different from a fortnight at 19 °C when it's plenty warm enough and a fortnight at 12 °C when you do need some heating. They're different HDDs at 15.5 °C base but the same at 20 °C base.

Posted By: atomicbisfI downloaded some but they appeared to be nonsense, but it was probably just because I didn't have an appropriate program to open them with.

I had a quick look and used 'ISOMERSE15', which seems to be Bath (there are other local ones). They seemed normal to me.

The 'Base Temperature' is always a hard one to decide on. 15.5°C is used as a standard as this allows you to compare two different buildings, does not tell you much about your particular building. As an example on iMeasure I used the lowest of 12°C and still got nonsense numbers. This may be because I don't turn my heating on till the outside temperature is at least 10°C or below. This is not really a refection of iMeasure, more a case that I live somewhere very mild, my house is well insulated and I am as tough, or as tight, as a Yorkshireman or a Cornishman (take your pick, a Cornishman is a Scott with all the generosity squeezed out of him).
Excel allows you to copy and paste text data in and then convert it to columns. It is under the Data tab and called Text to Column. You choose the 'delimiter' as 'comma' for a *.csv file. Make sure no other options are selected (like a full stop) as this may confuse the data.

Posted By: Ed DaviesBut - you then really need to do things at higher time resolution than per month down to of the order of the heating timeconstant of your building.

Yes, with any statistical based analysis, the higher the resolution the better the estimate. This is because the standard error of the mean (HDDs used to be set on the mean of two temperatures in the UK) needs a large number of datapoints to make up for the lack of daily ones.

As Ed Davies said earlier, other things can affect your space heating, solar gain is one but the biggest is windspeed (especially if you have a leaky house).
DHW is affected (or is if effected) by ground temperature, but this tends to be a gradual change over a week or two and can often seem out of step with air temperature. Ground temperature is affected by rainfall as much as anything else.

Posted By: skyewrightWhat ever is appropriate foryourbuilding.

Yes, but this then gives meaningless results when comparing against other places and other buildings.

I think the reason that the EST uses 15.5°C is because that is the mean air temperature that an average house in an average area needs to start heating. That in itself is pretty meaningless to us on here, but if this was stated on an EPC along side the HDDs, then it would become meaningful.

Paul
That may account for Canada's high energy usage then, or more likely the on/off nature of your weather. How many days does spring last, 3 isn't it

One half of Buffalo needed emergency aid while the other half needed a shovel last week didn't it?

I think you have summed it up, why I have never been happy with HDDs as a measure.
What it should do is show how effective any changes you make are.

It is really 'the curse of the mean' that is the problem. This tells you nothing about the distribution of temperature.
Generally temperature is evenly distributed with a relatively small standard distribution in the UK (have to be careful where you put the lowest and highest temperatures as they look different in the °C or K scale).
Then it all goes to pot when you take windspeed, direction and temperatures into account. I have higher temperatures the higher the windspeed, but I suspect that the NE coast of Scotland don't.

Posted By: Ed DaviesBut comparing buildings isn't the only use.

Isn't it more like that the HDD valueisthe result. Or, at least it's a big part of the result. Effectively, the regression line for a house is defined by its HDD base (the intercept) and energy required per degree day (the slope). Both matter.

Spot on. That's how we use it. It gives us a marker against which to measure changes in the house's performance as we make improvements.

How that compares to other houses isn't a particular concern for us. A body looking at things nationally, such as a government may have different priorities.

** Edited to reduce duplication. I wrote this one before the other (above) but it seemed to get lost when first posted. Now it has turned up!

Posted By: SteamyTeaThat may account for Canada's high energy usage then, or more likely the on/off nature of your weather. How many days does spring last, 3 isn't it" alt="" src="http:///forum114/extensions/Vanillacons/smilies/standard/bigsmile.gif" >

The high energy usage is because it's cold, not because of the basis temperature of HDD :)

What the basis temperature says is that if the external temperature is below the basis, the building will lose heat, if above, the building will gain heat. Now, your building could be losing heat when it's +18C outside, but the internal and incidental gains (insolation) could be greater than the heatloss at that temperature. HDD doesn't know or care about that - this is can only be computer by a proper heat load program that takes all these factors into account. The HDD basis temperature has to be the same for comparisons between two buildings at any two arbitrary locations to be made, but, of course, knowing the incidental gains also has to be taken into account as a location that's very sunny will lead to lower energy requirements than one that's dismal, everything else being equal - but only your heat load program will tell you whether your building will actually need heating (or cooling) at any particular external temperature.

For cooling, it's the same thing. A day with an 24 hour average of 18C is pretty warm - especially during the day time and is exacerbated by incidental gains so, again, the CDD figure won't tell you your cooling load.

As for Buffalo, that's an example of what a warm lake will do on a cold day. As the climate changes and the lakes get warmer, such lake-effect snows can only get worse - but more than six feet in a couple of days is pretty extreme!

Paul in Montreal.

Seret
You have just highlighted the things I don't like about them

They do need to be used with care, why I think they are not brilliant, but good enough for a first approximation.

They also go terribly wrong for people with storage heating (or broken thermostats) as this is a fixed load and is blind to temperature (not many people adjust the input and output for tomorrows temperature).

If people really want to get an idea of what is going on, then they have to start measuring. Then a real analysis can really show what is happening.
The equipment for this does not have to cost a lot (I do find it strange that people spend thousands on a heating system but nothing on monitoring).

When it comes to analysis, it is probably best to move away from HDDs and look at mean energy use and temperature differences (not hard if you have the data). You end up with a probability density for each temperature bin. Much more useful as it tends (a good stats term) to show when the heating is on or off with a step up or down in energy usage.

I am sure that when I first studied (as part of a BSc) HDDs we used Median rather than Mean. I think this goes back to when the Met Office collected data twice a day, 9AM and 9PM. The Median and the Mean is therefore the same (just the middle point of two). I think they then carried this practice on (and still do) as it gives a fair (another stats term) approximation.
Why I think the PDF (Probability Density Function) is easier and more enlightening.

Thanks Ed and Mr Tea, I've got some meaningful numbers now. I think the problem was I was using Open Office instead of Excell and its default settings for opening the file were wrong. The Oldfield Park station is surprisingly close - Mount Road is even closer but the microclimate is probably different. Mount Road is about 100 m higher and at the top of a hill open all the way to the Mendips to the south west so the wind blows almost constantly up there, even when it's calm down here.

How would you go about charting improvement in kWh/HDD? I was thinking of a separate trend line for each year. Or month on the X axis and kWh/HDD on the Y.

Ed

Posted By: atomicbisfHow would you go about charting improvement in kWh/HDD? I was thinking of a separate trend line for each year. Or month on the X axis and kWh/HDD on the Y.

Will have to think about that as HDDs are not really to do with the time of year, just time.

(Longish post, possibly boring)

Posted By: fostertomYes, the only way to use degree-days AFAIC, is to set base temp in fact equal to maintained internal air temp - 18C, 20C, 21C, whatever's desired

Yes that's the only way we will get meaningful national statistics which show the benefit of improving the building stock. TWh heating energy per 20C-HDD would be a meaningful national statistic. On the other hand...

Posted By: Ed DaviesIsn't it more like that the HDD valueisthe result. Or, at least it's a big part of the result. Effectively, the regression line for a house is defined by its HDD base (the intercept) and energy required per degree day (the slope). Both matter.

Ed, I do like this approach to monitoring an individual building, despite its snags.
The slope will be a property of the building, and possibly some local conditions such as wind velocities. The intercept (the HDD base) will also depend on the occupants, their stuff and their behaviour. If you take an empirical approach to measuring the building's thermal properties, you need to take a number of sample periods and plot the heating energy used, against the average temperature* for each period.
How long should the sample periods be? Periods shorter than the time-constant of the building won't tell you much. But long periods (weeks to months) won't give you many data points. And in the UK climate, these longer periods will almost certainly include warmer spells where no heating is needed - which will throw out the calculation.
Another snag is that the varying behaviour of the occupants will introduce uncertainty into the calculation of a value which should be fairly well-defined, i.e. the building's physical thermal properties. How much of the variation is random? You'll probably never fully comb out dependencies such as the greater use of lighting in winter.
We should be doing this for a lot of buildings, not just for the benefit of the occupants but to investigate how well buildings reproduce the thermal performance predicted mathematically from their design. Then we need to investigate whether the discrepancies are down to bad theory, or bad workmanship.

*SteamyTea has already highlighted the difficulties of averaging temperature. As ST indicates, we use should use average temperatures here, not HDD data, since we will be going around in circles if we presume an HDD base temperature.

A friend of mine has suggested using www.imeasure.org.uk and I've put the data I have available in. I have a mixture of quasi-monthly data from my previous supplier (generated from the fairly irregular readings I supplied online, ranging from a few days apart to about month apart, which their online thing used to estimate monthly usage) and actual meter readings from my current supplier (again fairly irregularly spaced). There is a gap of a month in between that I'm still crunching some numbers to try and fill.

Imeasure can do various things with the data. It's calculated degree day normalised 60 kWh/m2/year for 2014 (not sure if this accounts for the missing data for Jan and the rest of Dec yet to come), 110 kWh/m2/year for 2013 and 96 kWh/m2/year for 2012.

Ed