I’ve had a number of discussions recently with various people about the pros and cons of different options for Domestic Hot Water (DHW) pipework. You might not think there would be much to decide but in addition to choosing the pipework material (copper versus plastic) there are various options for routing the pipework around the house, with different pros and cons, and in particular there is a decision to make on the diameter of the pipes feeding the various fittings.
The (Main) Problem
Everyone has experienced the situation where you turn on a hot tap and you get cold water for what seems like ages before it finally starts to run warm and then hot. This is A Bad Thing because it wastes water, but it also wastes energy because once you turn off the tap the pipe is full of hot water (which you’ve paid to heat) which then cools down. Unless you use the same tap (or another one supplied by the same pipe) soon afterwards you have to go through the whole process again. Fitting low-flow taps to regular-sized pipes actually makes matters worse since you have to wait longer for all the cold water to flow out of the pipe before it gets replaced with hot water.
One solution to this problem is to install a pumped secondary circulation system, where the hot water gets pumped around a pipe loop which passes near to all the hot taps with just a short length of pipe connecting each tap to the loop. When you turn on a tap you get pretty much instant hot water, since only the short length of pipe needs to be flushed through; the water in the main loop is always* hot. (* OK, maybe not always, see later.) The problem with this approach is that while it saves water it wastes energy – you need to pay for the electricity for the pump, whenever that is running, and the pipe loop acts like a big radiator (even if it’s well insulated) which might be fine in the winter but is A Bad Thing in the summer, especially if you have a very highly insulated house which is in danger of overheating. It’s possible to mitigate both these effects by only running the pump when there’s a likelihood of someone wanting to use hot water, via either a time switch or motion detectors in bathrooms, and / or measuring the temperature of the water in the loop and only running the pump when it’s not already hot enough. That’s in danger of getting a bit complicated though, and while it might be the best option in a big hotel it’s almost certainly not optimal for a house.
As with a lot of things, it’s a good idea to go back to first principles and to try to find a simpler solution. Fundamentally, you need to run a hot tap until the cold water in the pipe between the storage tank and the tap has been replaced with hot water. The time taken to achieve this is governed by the volume of the pipework (length x area) and the flow rate of the tap. It’s always a good idea to minimise the length of the DHW pipework by central placement of the DHW storage tank and sensible routing of the pipes, but a big house is always going to have some quite long pipe runs so the other variable to watch is the diameter of the pipes. Taps with a maximum flow rate of 5 or 8 litres per minute don’t need large diameter pipes, and switching from a 15 mm pipe to 12 mm or even 10 mm can dramatically reduce the volume of water that needs to be flushed through. However, if the pipe is too small (especially if it’s quite long) that will restrict the flow so the right balance needs to be struck. The other problem with “micro-bore” pipes is that in hard water areas there’s less tolerance for them clogging up with limescale because there’s simply less space for the water to pass through.
Let’s address that water hardness question first. The water in most of the Midlands is supplied by Severn Trent Water who provide a Water Quality Search facility on their website. Along with other parameters this provides information on the hardness of the water, which for my location is classed as “Slightly Hard” with 10 Degrees Clark of hardness, equivalent to 8 Degrees German of hardness. (Interestingly my previous house, only 1 mile away, is supplied from a slightly different source and is a bit harder at 12.5 Degrees Clark.) From experience at my previous house limescale in pipes was nothing to worry about and the AECB Water Standards specify that water softeners should only be used where the supply is “Hard”, more than 14 Degrees Clark (and “Hard” > “Moderately Hard” > “Slightly Hard”).
Incidentally, dishwashers often need to be programmed with the local water hardness to control how often their built-in ion exchange water softener resin is regenerated using dishwasher salt. For water below 5 Degrees German, Miele dishwashers don’t bother softening the water.
I’m going to conclude that I don’t need to over-size the water pipes simply to prevent issues with them scaling up.
There are a few standards which address the length of time it takes for the water to run hot:
- The WRAS Water Regulations Guide specifies that 90 seconds is the longest you should have to wait
- This guide can be purchased from the WRAS website
- I need to do some more research to confirm this but I remember finding a table somewhere showing maximum pipe lengths for different pipe diameters, and a 15mm pipe has a 12m length limit.
- The AECB Water Standards specify a 1.5 litre “dead leg” (or 0.85 litre for their “Best Practice” standard). There’s a handy table in the AECB Water Standards which shows that a 10 mm PEX pipe would have to be 43.9 m long to contain 1.5 litres, and it’s impossible to imagine needing that length of pipe to reach any of the taps in my house.
- My intention is to at least achieve the AECB “Good Practice” standard, aiming for “Best Practice” in some areas.
- The UK Building Regulations (Approved Document G) includes a simple guideline:
3.7 Pipework should be designed and installed in such a way as to minimise the transfer time between the hot water storage system and hot water outlets.
The Water Supply (Water Fittings) Regulations 1999 cover various aspects of water installations, including hot water systems, but they’re primarily concerned with safety (e.g. preventing contamination and avoiding excessive pressure build-up) and do not address dead-leg volumes.
There are some excellent resources on the web which address various aspects of hot water pipe sizing:
- John Cantor’s “Heat Pumps” website has a page on Domestic Hot Water which covers pipe size in some detail and in particular covers the differences between copper and plastic pipe sizes which can have a dramatic effect at small pipe diameters,
- John Hearfield’s “Water Flowing in Pipes” pages cover the theory and practice of predicting flow and pressure in wonderful detail. The worked examples only cover copper pipes but the equations would be applicable to plastic pipes once adjustments have been made for the different internal diameters.
- While the equations look a bit scary at first glance, they’re able to cope with any liquid at any flow rate in any size of pipe. When using them in a DHW context things actually not too hard to follow – see below.
- Pipelife Ireland have some Qual-PEX Technical Literature (PDF link) which includes a pipe size / flow / pressure lookup page.
People like @AR_Clarke have done a lot of work on this topic too.
— Nick Grant (@ecominimalnick) September 23, 2015
Target Design Parameters
Based on all the above I’ve arrived at a set of design parameters for my house:
- I have good mains water pressure (seems to vary between 4 and 5 bars, normally about 4.5). While the water meter is a standard 15mm size the (long!) supply pipe from the meter is 32mm so that pressure will be maintained even when there’s a significant flow. Within the house the pressure will be restricted down to 3 bars using a pressure regulator in accordance with the AECB Water Standards and that regulated pressure should be maintained even when multiple outlets are open.
- It’s important to keep a reasonable pressure at the tap, even at full flow. Most fittings need at least 1 bar to work properly and deliver their specified flow rate at 3 bars. It’s unavoidable to have some pressure loss and there’s a need to compromise between the pressure loss and the pipe dead-leg volume. Permitting up to 1 bar of pressure loss seems reasonable, since that preserves at least 2 bars at the tap.
- I’d like to aim for a maximum 1 litre dead-leg volume, which is slightly more than the AECB Best Practice standard but still small by most standards.
- I’m going to be using water efficient fittings. In all cases I’ll comply with the “maximum consumption” limits of the 2015 Building Regulations and the AECB Water Standards. The fittings I’ve selected are:
- Basin Taps: 5 litres / minute
- Sink Taps: 6 litres / minute
- Actually I’m now favouring one with a nominal flow rate of 10 litres / minute, which is above the Building Regulations fittings limit, and would need a flow restrictor of some sort
- Showers: 9 litres / minute
- In addition to flow rate and pressure, the other consideration is noise from the water flowing through the pipe. This depends directly on the velocity of the flow, with 2 metres per second being a sensible upper limit.
10mm PEX pipe is widely available and will have a much smaller “dead leg” volume than the more conventional 15mm pipe. The question is whether 10mm pipe will restrict the flow too much, or be too noisy. Let’s do the sums using the Darcy-Weisbach equation by following the procedure from John Hearfield’s page:
- The first step is to calculate the Reynolds Number = Velocity x Diameter x (Density / Viscosity)
- For hot water at 50º C, Density / Viscosity = 2,330,000
- From the table on John Hearfield’s theory page
- The velocity derives from the flow rate and the pipe size. For 6 litres / minute through 10mm pipe (with a 7mm internal diameter) it works out at 2.6 metres / second.
- That’s well over the recommended limit for quiet operation (2 metres / second)
- Diameter is simply the 7mm, converted to metres
- The Reynolds Number (Re) works out at 42406
- For hot water at 50º C, Density / Viscosity = 2,330,000
- Knowing the Reynolds Number we can look up the Friction Factor, which is 0.022
- For a 10m pipe, the pressure loss works out at 1.055 bar
The good news is the calculation matches the table from Alan Clarke (9.5m limit for 10mm plastic pipe at 6 litres / minute) which gives me the confidence I didn’t do anything stupid; the bad news is that it’s going to be tough to route the pipework from the second floor to the ground floor in less than a 10m run, and the velocity is too high (so risks being too noisy). There are several pipe runs which will be over 13m, so it looks like 10mm PEX pipe is not really suitable.
OK, so what about 15mm PEX which is the next largest widely-available size?
- 6 litres / minute through a pipe with an internal diameter of 11mm (assuming a 2mm wall thickness) gives a velocity of 1.05 metres / second – well within the limit for noise.
- The Reynolds Number is 26935 so the Friction Factor is 0.025.
- For a 10m pipe, the pressure loss is only 0.125 bar.
- The problem is the “dead leg” volume which is 0.95 litres for a 10m pipe, and proportionately more for a longer pipe.
My conclusion is that 10mm PEX pipe isn’t likely to be suitable for the majority of the pipe runs, because of the pressure drop, and in any case it runs the risk of being too noisy at any flow greater than 4 litres / minute.
15mm PEX addresses the pressure drop and noise issues at the expense of dramatically increasing the time taken for the water to run hot.
12mm PEX pipe would be very suitable for the 5 – 6 litres / minute flow range typical of sinks and basins, but that doesn’t seem to be as widely available as the other sizes. Time to try to hunt some down…
The other main conclusion is that the different fittings are going to require different sizes of pipes – most likely distinguishing between basins, showers and baths.