The wine shed - a concept, and some horrible maths..
Posted: 07:48 Sun 04 Feb 2018
This is a long post – so please refresh your glass first..
One of the great revelations of my youth was the realisation that not only could reference books be wrong, but also that the authors of reference books copied each other’s mistakes.
Until a definitive work was published in the 1970s, countless authors had repeated the myth that Oak trees were slow growing, when in fact their dominance on dry clay soils in England is a consequence of the exact opposite.
Most people can remember the old Popeye cartoons and the tins of spinach – a throwback to a very old mis-analysis of the vegetable that led to the belief that it was exceptionally rich in iron. In fact it contains no more iron than other leaf vegetables.
Even today the myth is repeated that the Black Death was spread by rats, despite the abundant evidence that it spread across Europe at around 3-4 miles per day – a pace of movement entirely at odds with the habits of a rodent that likes to return to its nest each night.
So what about the oft repeated mantra that wine should be stored at a steady 10C/50F – who, I wonder, was the first person to put that notion into print?
Aside from being a nice round number on both temperature scales, it is clear that such conditions virtually never occurred before the days of industrial refrigeration.
The British Geological Survey states that seasonal temperature changes descend about 15m into the soil – or nearly six storeys down. This in turn is backed up by the Met office data for soil temperatures at 1m depth in the south of England, which show a consistent seasonal curve ranging from about 7C to 17C – so little attenuation at that depth of the surface temperature averages.
http://www.bgs.ac.uk/reference/gshp/gshp_report.html
https://www.researchgate.net/figure/225 ... th-East-UK
This data casts doubt on whether sub surface wine storage is actually worth the hassle, so it seems worthwhile to ponder the feasibility of constructing a surface ‘wine shed’.
So first, what are the actual needs?
10C is the year round average temperature in parts of central and eastern England, but in Bordeaux it is a full 3.5C warmer, and reaches nearly 15C in Porto.
It seems likely then that in the deepest and grandest Bordeaux château cellars, the temperature may never drop as low as 10C – yet it doesn’t seem to do any harm.
My own home observations also confirm that seasonal fluctuation is not an issue – the cellar under my home cycles through a range of about 9C summer to winter. Ullage measurements of sound mature bottles indicate a typical weight loss of around 100-200mg p.a.
Although neck profiles vary, the difference in weight terms between an IN level and a VTS level ranges from about 15 to 25g; so at the loss rates I am recording, it could take two centuries or more to achieve that degree of ullage. That’s good enough for me.
So it seems pretty safe to conclude that in moderation, summer/winter temperature fluctuations really aren’t a problem.
There are of course two other sets of fluctuations:
First the day/night fluctuations; these can range from virtually zero in overcast breezy damp weather to 10C or more when its dry, clear and still. Then there are intra seasonal fluctuations, such as a heat wave or cold snap. These happen constantly throughout the year to varying degrees, but in the UK, the 24hr average doesn’t often deviate by more than 5C from the seasonal average.
It is common currency that the first of these should be avoided at all costs, although the evidence to support that is a bit flimsy. My own observations suggest that problems arise when bottles are stored in places that see these day/night changes amplified, such as attics, vehicles and freight containers.
When stored in places that attenuate these daily temperature swings, such as below stairs cupboards, there is no obvious problem.
However, a design that eliminates day/night changes and moderates intra seasonal swings, seems sensible and sufficient, for the UK at least.
To combat the first of these fluctuations in our wine shed without using chillers requires a combination of insulation and thermal mass.
Every substance has a thermal conductivity, measured as watts per cubic metre when there is a one degree difference in temperature across the block.
This varies considerably – an insulator like expanded polystyrene has a thermal conductivity of 0.038 whereas a dense concrete block has a conductivity of 1.13
Thermal mass on the other hand, is derived from the specific heat of a substance.
The specific heat capacity of a substance is the amount of energy needed to raise the temperature of one kilogram by one degree. To derive the thermal mass this then has to be multiplied by the density of the substance.
Thus the thermal mass of expanded polystyrene (as EPS70) is it’s density of 15kg/m3 times its specific heat 1400J/Kg/C making a thermal mass 21000J/m3/C
Dense concrete on the other hand is around 2000Kg/m3 with a specific heat of around 1000, resulting in a thermal mass of 2000,000J/m3/C – or nearly 100 times more than expanded polystyrene.
The thermal mass matters because when heat is applied to one side of a brick or block, it does not immediately manifest itself on the opposite face. The block itself has to warm up first to create a thermal gradient before any heat transfers.
Calculating this transfer time is significant, since if the walls of our wine shed take more than 12 hours to warm up, the day/night temperature fluctuations will cancel each other out with only the faintest temperature ripples apparent on the inside face.
The maths here is a bit counter-intuitive, and as I’ve not been able to find any clever formulae to assist me, I’d be grateful if others would scrutinise my calculations. I also thought someone would have given this warming up period a clever name, but as I’ve drawn a blank on that as well, I’ll call it hysteresis.
Getting my head round the calculation for this was harder than I expected.
The first thing I deduced was that as the amount of energy needed rises in proportion to the temperature differential, so the amount of time needed for hysteresis will remain the same, irrespective of the temperature difference.
The second curiosity is that the progress rate of heat through a block as it forms a temperature gradient does not appear to be linear – it appears to start very quickly, slow towards the middle and then speed up again towards the end.
Less surprising was that the calculation gives an exponential result. The hysteresis time quadruples as the thickness doubles.
The only way I could make sense of the calculation was to consider the subject as a succession of 1mm thick slices, and using a spreadsheet calculate the time each successive slice would take to reach the requisite temperature, given the distance from the starting face, and the fact that the thermal gradient demands less and less energy for each successive slice.
I worked on the assumption that the source of heat would be absolute to give a worst case figure, and then totalled the times for each of the slices.
Does this give me the correct answer? I’m bugged that I might have missed something..
Having drafted a spreadsheet to compute this, it was only a simple matter to then compute the thickness of various materials needed to achieve the 12 hour hysteresis target.
If you haven’t already descended into a comatose state, this is where it gets interesting – well I think so, anyway..
If your walls are made of dense concrete, the thickness needed for a 12 hour hysteresis computes as 383mm, but if they are polystyrene that figure rises to 685mm. Despite far superior insulating properties, heat will emerge from the opposing face of a polystyrene slab far sooner than a concrete one of the same thickness, although the amount of heat passed by the concrete will be very much greater once the hysteresis period is over.
Now consider the properties of an aerated concrete block, commonly known as Breeze or Celcon blocks in the UK, Cinder blocks in the US. These lightweight blocks combine a moderate amount of thermal mass (just over 30% that of dense concrete) with moderately good insulating properties (polystyrene is four times better)
Putting the data on my spreadsheet revealed a 12hr hysteresis thickness of just 249mm.
One of the standard sizes for these blocks (in the UK) is a thickness of 275mm. The blocks are inexpensive, light, easy to cut and very quick to lay.
It seems ideal for eliminating the day/night element. However, before I get carried away working out a cost effective system for muting the intra seasonal fluctuations, and address issues like floors, ceilings and doors, I would be very grateful if those whose maths is not quite as rusty as mine would check my calculations.
The essential data for these blocks (taken from the Celcon data sheet for their standard block) is:
Density – 600Kg/m3
Specific heat capacity – 1050J/KgK
Thermal conductivity – 0.15JmK
One of the great revelations of my youth was the realisation that not only could reference books be wrong, but also that the authors of reference books copied each other’s mistakes.
Until a definitive work was published in the 1970s, countless authors had repeated the myth that Oak trees were slow growing, when in fact their dominance on dry clay soils in England is a consequence of the exact opposite.
Most people can remember the old Popeye cartoons and the tins of spinach – a throwback to a very old mis-analysis of the vegetable that led to the belief that it was exceptionally rich in iron. In fact it contains no more iron than other leaf vegetables.
Even today the myth is repeated that the Black Death was spread by rats, despite the abundant evidence that it spread across Europe at around 3-4 miles per day – a pace of movement entirely at odds with the habits of a rodent that likes to return to its nest each night.
So what about the oft repeated mantra that wine should be stored at a steady 10C/50F – who, I wonder, was the first person to put that notion into print?
Aside from being a nice round number on both temperature scales, it is clear that such conditions virtually never occurred before the days of industrial refrigeration.
The British Geological Survey states that seasonal temperature changes descend about 15m into the soil – or nearly six storeys down. This in turn is backed up by the Met office data for soil temperatures at 1m depth in the south of England, which show a consistent seasonal curve ranging from about 7C to 17C – so little attenuation at that depth of the surface temperature averages.
http://www.bgs.ac.uk/reference/gshp/gshp_report.html
https://www.researchgate.net/figure/225 ... th-East-UK
This data casts doubt on whether sub surface wine storage is actually worth the hassle, so it seems worthwhile to ponder the feasibility of constructing a surface ‘wine shed’.
So first, what are the actual needs?
10C is the year round average temperature in parts of central and eastern England, but in Bordeaux it is a full 3.5C warmer, and reaches nearly 15C in Porto.
It seems likely then that in the deepest and grandest Bordeaux château cellars, the temperature may never drop as low as 10C – yet it doesn’t seem to do any harm.
My own home observations also confirm that seasonal fluctuation is not an issue – the cellar under my home cycles through a range of about 9C summer to winter. Ullage measurements of sound mature bottles indicate a typical weight loss of around 100-200mg p.a.
Although neck profiles vary, the difference in weight terms between an IN level and a VTS level ranges from about 15 to 25g; so at the loss rates I am recording, it could take two centuries or more to achieve that degree of ullage. That’s good enough for me.
So it seems pretty safe to conclude that in moderation, summer/winter temperature fluctuations really aren’t a problem.
There are of course two other sets of fluctuations:
First the day/night fluctuations; these can range from virtually zero in overcast breezy damp weather to 10C or more when its dry, clear and still. Then there are intra seasonal fluctuations, such as a heat wave or cold snap. These happen constantly throughout the year to varying degrees, but in the UK, the 24hr average doesn’t often deviate by more than 5C from the seasonal average.
It is common currency that the first of these should be avoided at all costs, although the evidence to support that is a bit flimsy. My own observations suggest that problems arise when bottles are stored in places that see these day/night changes amplified, such as attics, vehicles and freight containers.
When stored in places that attenuate these daily temperature swings, such as below stairs cupboards, there is no obvious problem.
However, a design that eliminates day/night changes and moderates intra seasonal swings, seems sensible and sufficient, for the UK at least.
To combat the first of these fluctuations in our wine shed without using chillers requires a combination of insulation and thermal mass.
Every substance has a thermal conductivity, measured as watts per cubic metre when there is a one degree difference in temperature across the block.
This varies considerably – an insulator like expanded polystyrene has a thermal conductivity of 0.038 whereas a dense concrete block has a conductivity of 1.13
Thermal mass on the other hand, is derived from the specific heat of a substance.
The specific heat capacity of a substance is the amount of energy needed to raise the temperature of one kilogram by one degree. To derive the thermal mass this then has to be multiplied by the density of the substance.
Thus the thermal mass of expanded polystyrene (as EPS70) is it’s density of 15kg/m3 times its specific heat 1400J/Kg/C making a thermal mass 21000J/m3/C
Dense concrete on the other hand is around 2000Kg/m3 with a specific heat of around 1000, resulting in a thermal mass of 2000,000J/m3/C – or nearly 100 times more than expanded polystyrene.
The thermal mass matters because when heat is applied to one side of a brick or block, it does not immediately manifest itself on the opposite face. The block itself has to warm up first to create a thermal gradient before any heat transfers.
Calculating this transfer time is significant, since if the walls of our wine shed take more than 12 hours to warm up, the day/night temperature fluctuations will cancel each other out with only the faintest temperature ripples apparent on the inside face.
The maths here is a bit counter-intuitive, and as I’ve not been able to find any clever formulae to assist me, I’d be grateful if others would scrutinise my calculations. I also thought someone would have given this warming up period a clever name, but as I’ve drawn a blank on that as well, I’ll call it hysteresis.
Getting my head round the calculation for this was harder than I expected.
The first thing I deduced was that as the amount of energy needed rises in proportion to the temperature differential, so the amount of time needed for hysteresis will remain the same, irrespective of the temperature difference.
The second curiosity is that the progress rate of heat through a block as it forms a temperature gradient does not appear to be linear – it appears to start very quickly, slow towards the middle and then speed up again towards the end.
Less surprising was that the calculation gives an exponential result. The hysteresis time quadruples as the thickness doubles.
The only way I could make sense of the calculation was to consider the subject as a succession of 1mm thick slices, and using a spreadsheet calculate the time each successive slice would take to reach the requisite temperature, given the distance from the starting face, and the fact that the thermal gradient demands less and less energy for each successive slice.
I worked on the assumption that the source of heat would be absolute to give a worst case figure, and then totalled the times for each of the slices.
Does this give me the correct answer? I’m bugged that I might have missed something..
Having drafted a spreadsheet to compute this, it was only a simple matter to then compute the thickness of various materials needed to achieve the 12 hour hysteresis target.
If you haven’t already descended into a comatose state, this is where it gets interesting – well I think so, anyway..
If your walls are made of dense concrete, the thickness needed for a 12 hour hysteresis computes as 383mm, but if they are polystyrene that figure rises to 685mm. Despite far superior insulating properties, heat will emerge from the opposing face of a polystyrene slab far sooner than a concrete one of the same thickness, although the amount of heat passed by the concrete will be very much greater once the hysteresis period is over.
Now consider the properties of an aerated concrete block, commonly known as Breeze or Celcon blocks in the UK, Cinder blocks in the US. These lightweight blocks combine a moderate amount of thermal mass (just over 30% that of dense concrete) with moderately good insulating properties (polystyrene is four times better)
Putting the data on my spreadsheet revealed a 12hr hysteresis thickness of just 249mm.
One of the standard sizes for these blocks (in the UK) is a thickness of 275mm. The blocks are inexpensive, light, easy to cut and very quick to lay.
It seems ideal for eliminating the day/night element. However, before I get carried away working out a cost effective system for muting the intra seasonal fluctuations, and address issues like floors, ceilings and doors, I would be very grateful if those whose maths is not quite as rusty as mine would check my calculations.
The essential data for these blocks (taken from the Celcon data sheet for their standard block) is:
Density – 600Kg/m3
Specific heat capacity – 1050J/KgK
Thermal conductivity – 0.15JmK