The acoustics of air source heat pumps - part 2: How loud are heat pumps, anyway?

Most of the time, acousticians have no choice but to rely on the information that heat pump manufacturers include on product data sheets. Exactly what form this takes can vary quite a lot, but it usually gives a single sound level, either broken down by frequency or as a single-number “dBA” level.

The numbers on these datasheet are the result of laboratory tests, following standards which define the conditions the heat pumps should be operating under when the noise levels are measured. 

We don’t have very much information at all about how the sound it makes in the laboratory test chamber relates to the sound it will produce once it’s installed at someone's home and providing heating and hot water. We decided to do something about that.

We set up a sound level meter at one meter from an ordinary air source heat pump in the garden of one of our MEP colleagues over this last Christmas break, covering a period of just over two weeks. 

We chose this time of year to measure the heat pump doing as much as possible; not only was it cold (we even had snow), but people would be home more than usual and using plenty of hot water to shower and do all the washing up after Christmas dinner.

As well as our noise measurements, we also had monitoring of the heat pump and heating system set up, which gave us access to a lot of extra data about what the heat pump was doing. This included:

• Electrical energy input
• Heat energy output
• Flow temperature (how hot the water was as it left the heat pump)
• Return temperature (how hot the water was as it came back to the heat pump going round the house)
• The outside temperature
• The temperature inside the house

The owner of the heat pump also agreed to try a number of different combinations of settings across the period, to see what the impact of these settings have on the noise produced by the unit.

All of these measurements tell us about one particular heat pump, attached to one particular heat system, in one particular house. There are plenty of reasons that what we measured here isn’t broadly applicable to air source heat pumps in general.

The house this heat pump serves is a relatively small mid-terrace house with better than average insulation, so a bigger house with a bigger heating load might look different. 

Also, this unit is owned and operated by an MEP engineer, who was quite involved in the design of the system, and does a lot of ongoing work to get the best efficiency as possible out of his unit.

So that we’ve got something to compare against, we used the manufacturer’s sound power data to predict what we’d expect the level at one meter to be, using a standard spherical sound propagation with two reflecting surfaces (the ground and the solid fence between the unit).

These predicted levels are the sort of thing we’d normally have to rely on for an assessment of a new unit that we only had the datasheet for.

The dotted lines on the graphs below show the predicted levels for heating and hot water, based on the different water temperatures we expect for the two types of operation. The solid lines show the measured noise levels, with the orange lines showing the times the heat pump is in heating operation, the purple line showing the hot water operation, and the grey lines showing the times the heat pump isn’t programmed to be on.

The first thing that’s obvious from the measured noise levels is that noise from the heat pump is far from being a constant, steady noise source. For good portion of the day, the heat pump is on standby, as it’s only programmed to provide heating for a few hours per day, and hot water for one burst of less than an hour each day.

When it’s providing heating (shown in orange), you can see that the level does peak at something close to the level we’d predict from the datasheet, but then it drops down to a noise level 3-8dB below. It is at this lower noise level much more often than it is at the peak level.

The hot water cycle (shown in purple) is a lot shorter, and stays at a fairly constant level until it turns off. The level it stays at is pretty similar to the predicted level, as well as being pretty similar to the peak level during the heating cycle. 

The graphs below show the measured noise levels day-by-day.

When it comes to the different settings, we can see that the noise reduction mode operates as you’d expect from the name. The left-hand graph here covers the first three days, where the noise reduction mode was off, and the right-hand graph shows the next three days where the noise reduction mode was on. If you compare them, you’ll see that the noise levels are lower by around 2dB.

Another thing we can see fairly easily is that the external temperature has an effect of the noise levels the heat pump produces.

In the two graphs below, the settings relating to the heating cycle are exactly the same. All that’s different is the external temperature. 

The left-hand graph shows that for three days and with relatively consistent outside temperatures, for the most part, the noise levels are pretty similar each day. The right-hand graph shows another three days, over which time the temperature drops pretty sharply, and you can see that the noise levels go up sharply too.

In fact, the 2^nd of January, the last day on that second graph, sees the lowest outside temperature and the highest levels of heat pump noise measured across the whole period.

There are a number of elements which will be discussed in more detail including the relationship between heat pump efficiency and noise levels and what causes the spike in noise when the heating system first turns on.

For now though, the main takeaway is that a noise level measured by the manufacturer in laboratory conditions does not give the whole picture. The noise produced by a heat pump in real-life usage is highly variable - depending not just on the make and model of the heat pump but also the heating system it's connected to, the way it’s controlled, and the weather conditions.