Thursday, 6 March 2014

Getting the best from underfloor heating



Many years ago, when heat pumps were not so common, I found it a real struggle to get any of the main underfloor heating suppliers to embrace the need for low water temperatures for use with heat pumps.
15 years on, and I’m sure that things have changed dramatically, but I’m still stumbling across things that make me doubt that.
When I visited Germany over 10 years ago, I got the feeling they commonly use much closer spacing between floor pipes than we do in the UK, and I have read of very close spacing (50mm) in Austria.  I realise these are potentially cold countries, but even recently visiting a Spanish heat pump company, they were surprised that we don’t use closer pipe spacing (more pipes).

Over the last 6 months I have heard several references in the trade to; ‘never use closer than 150mm pipe spacing’.   However, CIBSE clearly give ratings for 100mm spacing, as does the MCS Emitter Guide, so why are some in the UK so resistant to putting more pipe into floors.

To basics  -   COP v heat pump water temperature
Heat pumps are more energy-efficient if the water temperature is low.
Coefficient of performance (COP) = heat output / electrical power input

This graph of an air source heat pump shows how beneficial a low flow temperature is.   Units like this with electronic expansion valves are particularly efficient at very low water temperatures.

How to get a low flow temperature with underfloor heating

The values for the example graph above have been extracted from CIBSE, and show a general trend for a fixed heat output.  It demonstrates what we all should know  - the closer the spacing, the lower the required water temperature for the same heat output.
Putting the two together we could plot pipe spacing v COP.  It would clearly indicate that closer spacing improves COP, so why do some seem reticent to embrace close spacing?  Is that added cost for more pipe really that much??
I have gleaned that some people have the notion that close spacing could lead to warm floors and an overly hot house.  The whole point here is that with closer spacing, we can turn the water temperature setting (heating curve) down, and get the same heat output, thus improving the COP.
I think some of the fears about hot  floors stem from boiler systems, many of which had plenty of extra output capacity for quick warm-up, and hot patches could result. However, heat pumps are generally slower response, with lower water temperatures, so are much more forgiving in this respect.  I recently visited a system where I had used 50mm spacing in an always-open loop in a bathroom.  I asked the owner if they had ever thought the bathroom floor was too warm.  The answer was no, never. 
The reason why I had used 50mm spacing was that we didn’t want a buffer cylinder, so I was trying to ensure there was adequate water quantity, and flow-rate in the floor – effectively using the floor as a buffer.  
Another hot (excuse the pun) topic here is floor coverings. People are now seeing that a carpet will drop the star rating for RHI, so there is an added financial incentive for having solid (tiled etc.) floor, as seems far more common in Germany, instead of carpet.
However, by designing a house where some rooms are tiled, and some are carpet, you are somewhat limiting any changes that future occupants might wish.  (e.g.  if 100mm spacing in a carpeted room, and 200mm in a tiled room).   Underfloor pipes are literally set in stone, so can never be changed.  The only solution that I could think of here would be to interlace the loops (given that there is often more than 1 loop per room).  There is always the option of turning off one of the interlaced loops.  I like multiple interlaced loops.
I have come across an installation where a new owner had fitted thick pile carpet in one room without thinking the affect it would have.  The result was a cold room, remedied only by turning up the water temperature, thus  increasing the running cost, and the reliance on room thermostats to limit heat to the tiled rooms.  
The general design approach for underfloor is to consider the heat required (e.g. watts/sq m) , but unless a buffer cylinder is fitted, we should also consider what heat is being delivered to the floor, given that some zones will be closed for some of the time, and the outside temperature is seldom at design temperature (-2C etc.).   For most of the time, we have far more available heat than the floor needs.  Even with modulating heat pumps, there can still be a tendency for the flow temperature to stray above the theoretical flow/return temperatures. This is another reason for favouring more pipe in the floor.  Furthermore,  MCS requires the heat pump to provide at least 100% of the design at -2C etc.  Due to models only being available in certain size jumps, the heat pump installed is often oversized, so this is an added reason for ensuring that there is adequate pipe in the floor.
For any thinking that buffer cylinders are the perfect answer – they may be an answer, but they are seldom perfect, as shown by this piece of monitoring.     (openenergymonitor.org)

In this example of a system with a simple buffer tank, the heat pump flow/return needs to be approx. 4 degrees hotter than UH flow/return.    This could reduce energy efficiency by 10%, plus the added energy to run a second circulation pump.
A buffer can usually be avoided IF enough zones are always on AND if there is plenty of pipe in the screed.
Another potential worry is the thermal mass of the screed.   I was recently involved in the design of a passive house where a buffer cylinder was not wanted, so I proposed  to use the screed as the buffer, and fitted 9 x 100m loops with average 100mm spacing in a  floor that was 200mm thick.  One might expect some temperature overshoot, and I may not have been so bold as to propose it if it were not for the fact that due to the interlaced nature of pipework,  we could shut off ½ the zones if we needed to.
The result was surprisingly good with all room zones open and control on one master thermostat. The whole house has very even temperatures.  This demonstrates the self-regulating nature of very low temperature underfloor heating.
Another concern I have heard of is the pumping power required  for such a lot of pipe.  In the above passive house example.  If the heat pump needs 10 lit/min to give a 5 degree flow-return dt , then the flow rate is about 1.11 litres/min for each pipe loop.  If we used a more standard 5 loops, the flow rate per loop would be 2 litres/ min.  For the same heat (kW) and same dt, more pipe actually means less pumping power since the heat output per m of pipe is lower, and the flow-rate for each loop  is lower.
Finally, I have also come across the notion that mixer valves and pumps are desirable, even for a heat pump.  Here is an example to illustrate the potential penalty of having a mixer even if the mixer never actually mixes (i.e. its fully open).

The arrangement here with recorded temperatures shows that the flow to the floor is always equal in temperature to the return to the heat pump. In this example, the flow from the heat pump is 7 degrees higher  than the flow to the floor.  The mean floor temperature  was only 23°C (average of 28 & 18).
If the mixing valve and pump were removed, the heat pump’s flow could go directly to the bottom manifold.  
To get the same floor heat output, the heat pump setting could be adjusted down so as to give a working flow of 26 and return of 20.  This is a reduction in flow temperature of 9 degrees, potentially (according to our first graph at top of page ) saving 13%.  
(see top graph - The COP at 35°C is 3.1, at 26C its 3.1, saving 13% in COP). 

All these little details can eat into potential savings. They should be dealt with at design stage.

Thursday, 20 February 2014

Potential perils of plastic pipe




A college tutor recently suggested to me that the current environment in the heating industry does not encourage thinking.   Maybe I should therefore be less amazed when I come across plumbers who think copper and plastic pipe are interchangeable. 

Environmental considerations aside, a more direct issue is that of flow rate and pressure drops.  Bore size of plastic is considerably smaller than copper, and it seems that this is often overlooked.  I am not against plastic, but if installers select plastic pipe instead of copper without checking the size correctly, this could have a negative effect on heat pump’s COP.  

(If you drop-off ½ way through this - read the scenario at the end )


With respect to the materials of plastic and copper, to my surprise I found a very informative document that discusses primary energy of these two materials in some detail, and concludes that the total energy involved in mining and manufacturing copper is far greater than the total energy (including the crude oil) to manufacture plastic pipe.  http://www.hepstore.co.uk/downloadPDF.aspx?id=840  Looking at other potentially less biased general data on copper and plastic, it seems to confirm that more oil is used processing copper than would be used to produce plastic.  

Another factor to consider, that mainly affects pipe runs to hot taps etc, is the heat capacity of the pipe material. Plastic has a relatively high specific heat, and the wall is thicker, but it’s light. The net thermal capacity of the two is fairly similar.
(The cold feel of copper is more to do with conductivity from the hand than heat capacity)

However, obvious factors aside, one of the biggest issues that could affect installations involving heat pumps relates to the internal bore diameter.  This could have a noticeable effect on the energy-efficiency of the system.

All metric pipes are measured by their outside diameter. As can be seen, with common pipe sizes (outside diameters), equivalent plastic pipes have considerably smaller internal area to copper. This has a dramatic effect on flow characteristics.

 The graphs below illustrate the relative internal dimensions of common pipes.

 (see AECB Water Standard)




A brief note about smoothness - It seems a common belief that plastic is ‘smoother’ than copper, but the inner wall ‘smoothess’ of the two is the same.  However, plastic can be one-piece with slow sweep bends. This is certainly ‘smoother’ than copper with tight elbows.   Re inner surface, we can assume the two materials are the same.

Whilst it is fairly easy to look-up the pressure drop resulting from a specific flow rate with a specific pipe, we can see from the 2nd graph at-a-glance the relative flow capacity since the cross sectional area loosely indicates flow capacity.

If a certain flow rate is required, then we can look-up the required pressure that is required across the pipe length (beginning of pipe to end of pipe length).  The internal bore must be chosen such that the circulation pump is not overly large and energy-wasteful.

This graph shows the pressure required to maintain a certain flow rate for a fixed length of pipe.  As can be seen, the pressure drop along the example pipe varies very dramatically, so the wall thickness makes very big difference. 
In this example, we can see that a 15mm copper pipe could be used with a common central heating pump (shown at 3.6m head, 36kPa). However, if plastic were chosen, then one would need almost 9m head to achieve the required flow - far beyond the capability of normal circulators. On the other hand, if 22mm plastic were chosen, the pressure requirements would be only 1m head (10kPa) which is likely to achieve very low circulation pump energy.

In real life, we tend to have a pump connected to a pipe system, and the flow rate that results is dictate by the balance between the pressure produced by the pump and the ‘restriction’ of the entire pipe work circuit.

For our final graph, we consider a pipe circulating with a fixed-pressure electronic pump (Alpha etc)


The above graphs show relative changes in flow rate that would result from a fixed pressure. If the pipe chosen were too small, then a larger circulation pump may be needed in an attempt to compensate for the extra restriction caused by the small internal bore.

The point here is that by choosing plastic instead of copper of the same nominal size, the system could potentially suffer unless the sizing is checked. There is of course no problem using plastic if it’s the right diameter.  Indeed, 28mm plastic may be an ideal choice for the connections from a heat pump simply to minimise noise transmission.  The best final solution is often a mixture of both plastic and copper for a multitude of reasons.

All this emphasises how dramatically the wall-thickness affects flow rates and pump pressure requirements, but how does this translate into reduced COPs?

Scenario (based on something I observed on a barn conversion)

Let us consider a radiator at a far distance from the heat pump.  The flow-rate relates to the pressure drop, which relates directly to the pipe-run length, and of course, the required flow-rate relates to the room size (bigger the heat demand, the more flow required).  In this instance the room is large.
The default pipe size choice would be normal 15mm (outside diameter), but if the sums are done, it may become apparent that the choice should be between either 15mm copper or 22mm plastic. 

How could a pipe with too small bore affect the COP? 

Radiators should be balanced, in general by throttling valves (lockshield)  on smaller radiators, and those with shorter pipe runs. However, this is actually quite a difficult thing to achieve with a heat pump because the temperature difference (water inlet to water outlet) may only be 5 degrees (°C).  (It’s much easier to measure and adjust when the temperature differences are larger).

It is not ideal, or easy, to have to throttle most other radiators on a system, and there is the added risk of the circulation pump not being big enough to cope.
The likely result of any ‘restrictive’ pipe-runs would be a reduced flow rate to the radiator.  This would result in a considerable area (the bottom) of the radiator being cool, and a reduced heat output to that room.
The obvious action to redress the short-fall of heat would be to increase the flow temperature by increasing the heating curve setting of the heat pump. i.e. increasing the flow temperate from say 40 to 45°C.    Now the heat pump has to heat ALL water to a 5 degrees higher level.  This is likely to reduce the COP by 10 to 12%.

This all indicates that one must never assume copper and plastic are interchangeable without considering the pressure drops and diameters.  That aside, we have also highlighted the importance radiator balancing.  Ideally the pipe runs to radiators would be laid out so that the flow rates are naturally about right without the need for much valve adjustment.  A little extra work on the design makes life much easier thereafter.

Anyone intersting in further reading on water flows - this site is very intersting
http://www.johnhearfield.com/Water/Water_in_pipes.htm