Noise Attenuation By Metal Roofs

Stuart Hayman, March 2012

In October 2011 we updated you on the next stage of the noise through roofs technical project. The Acoustics Research Group of the University of Canterbury did some research for NZMRM to determine the
attenuation of road traffic noise through various roof cladding materials and roof/ceiling installation
systems used in New Zealand. The research was designed to provide a comparison between these
various systems in order to determine the best and most cost-effective way to reduce the level of traffic
noise entering the living spaces via roof/ceiling system.

This article describes this project in more detail and discusses the way in which the work was done and the outcomes. We have recently completed a technical summary of the results, which together with the full reports from the University are now available on the NZMRM website,


The genesis of this work goes back to the Stonefields development in the Mount Wellington quarry in  Auckland.

The quarry site is bounded on all sides by roads which take reasonably high levels of traffic (more now that the development is proceeding). One side the road is in close proximity to the housing, however in most of the quarry development the housing is further away from the through roads or well below road level. Auckland City Council has a District Plan which requires noise levels in bedrooms to be no more than 35
dBA L10 between the hours of 10 pm and 7 am and no more than 45 dBA L10 in other rooms at any time.
When Fletcher Residential were looking at building many of the houses in the Stonefields development they wanted to make sure they complied with the District Plan. For the houses most exposed to road noise, they engaged one of the principal NZ acoustic consultants to advise them.

An initial investigation report was produced in 2006, followed up in 2008 as building commenced. This report concluded that for the most exposed houses the 35 dBA level in the bedroom (generally of course the nearest rooms to the roof) could be met by:

1. Long run metal roof cladding on 17.5 mm plywood with resiliently supported ceiling of 13 mm GIB
Noiseline® and sound absorptive material of R1.8 in the cavity; or

2. Textured metal tiles on 17.5mm plywood with resiliently supported ceiling of 13mm GIB® Standard
plasterboard and sound absorptive material of R1.8 in the cavity; or

3. Concrete tiles with resiliently supported ceiling of 13 mm GIB Noiseline® plasterboard and sound
absorptive material of R1.8 in the cavity.

So, ignoring the minor idiosyncrasies that metal tiles can use a GIB® Standard plasterboard and the other two require GIB Noiseline® plasterboard, and that the plywood was not 17 or 18 but 17.5 mm, this was all bad news for metal roofing, and almost before we got onto it, we had some long run roofs fitted on plywood in Stonefields. Of course this adds significantly to the cost. There are questions about how to fix roofing to sarking properly and the effect on moisture transmission and ventilation, both of which are important and complicated by the use of ply.

Different acoustic experts were consulted. They noted that the report quoted very little actual on-site noise measurement. We “borrowed” a new house on one of the affected roads which had a metal tile roof and installed sound measurement equipment inside and outside, and monitored the external and internal noise levels over an entire day/night. The results showed that the outside noise from the road was indeed around
the estimated level, but the level inside the bedroom was less than 35 dBA. This was enough to show
Fletcher Residential that in this particular part of the development, we didn’t need plywood under metal
tiles, but it didn’t resolve the wider issue and it wasn’t long before we (NZMRM) received reports of jobs
lost elsewhere to concrete tiles, because of this report. Houses elsewhere in Stonefields were built
with plywood under metal roofing or used concrete tiles.

We consulted further with people involved in building noise control, and it became obvious that to resolve the issue authoritatively we were going to have to undertake some research ourselves. For several reasons we engaged the Acoustic Research Group at the University of Canterbury to do this work for us.


The first step was to commission, with some funding from TechNZ, a literature survey of published work worldwide to see what else had been done and where. Our researcher at Canterbury discovered that while there has been a lot publication of intrabuilding noise control research, in fact very little work had been
published globally about our specific interests, transmission of road noise through roofing. Existing publications suggested other factors affected noise in rooms – the transmission paths through the windows, for example. It also showed that increasing the ceiling insulation had been effective in reducing noise transmission through roofs. Some work had been done on reducing aircraft noise through roofs, which is a source directly above the roof, and we discovered that in several places, ceiling insulation had been mandatorily increased under flight paths, including in New Zealand.

From this we concluded that we would need to do the research ourselves. With further valuable support from TechNZ, we developed a programme with the University of Canterbury to investigate the noise attenuation properties of roofing installations using the three main roofing materials used in NZ, to see
how attenuation could be improved by various means.

All the testing was carried out to international standards by a recognised authority and therefore these results can be relied upon.


Firstly we tested variants of the different roof claddings (various types and thicknesses of longrun roofing, painted and textured metal tiles, and concrete tiles with and without underlay). From this we determined what did and did not have an impact, and whittled the products down to three basic types:

  • Long run corrugate 0.4 mm (with underlay as is required) as representative of profiled metal roofing (longrun)
  • Pressed Metal tiles, textured (with underlay as is required) as representative of pressed metal tiles
  • Concrete tiles, with and without roof underlay (underlay is optional, but as you will see, important)

Then we took these three types and one variant and installed them in a standard roof installation, using the
current H1 requirements for ceiling insulation (which is higher than the 2006 and 2008 reports referenced).
Then we looked at the effect of the following variables (i.e. used when not normally used, or in increased

-> Plywood sarking
-> Roof underlay
->GIB® plasterboard ceilings

The results were somewhat unexpected and very interesting, and have proved, amongst other things, that adding plywood decking under roofing is the not the most effective way of reducing roof transmission noise, and is very expensive.


The test process looks only at the sound transmission through the different roof systems used, and makes no attempt to measure sound transmission through the road-facing wall of the building, or through the house structure. The method used was to create a 4.5 m wide x 2.4 m high hole in a solid concrete wall. This wall separates a “reverberation” room and a “semi-anechoic” room. (These two rooms simulate
the typical external and internal conditions). The roof material is mounted over this hole and the roof structure/ceiling is mounted in the room on one side. While this looks somewhat strange (the “roof” is vertical and the “ceiling” at the angle of roof pitch to vertical), it does simulate a roof/ceiling structure realistically.

A diffuse sound field was created on the cladding side of the roof using loudspeakers which generated
pink noise. The sound intensity on the ceiling side was measured to determine the sound reduction
index following the guidelines of the international standard, ISO 15186-1. The measurements were relatively quick, but the reconstruction of each set of roof structures takes much longer. So you might have
a day’s work setting up a roof and three hours doing the testing, then change the roof over again. The
actual sequence of test structures was set so as to minimise the building work.

So, what were the results?

Of course there are a huge number of individual test results but each lot of testing has been illustrated in several graphs, and these are presented in the Acoustic Research Group’s reports on the NZMRM website.

These graphs show the sound reduction index for the various materials/installations at the 1/3rd octave frequency bands. The higher the lines are on the graph the better. The basic test method complies with ISO 15186-1:2000 which covers frequencies from 100 to 5000 Hz.

The roof structures have significantly higher noise attenuation than the roof claddings – as you might expect.

What you might expect is that the air-tightness of the longrun roofing would be effective in reducing
higher frequency noise, and that metal tiles would be worse and concrete even worse. Concrete might be expected to reduce low frequency noise. Some of this is true, but the rest is not so obvious.

The full technical report is finished, and is now on the website but here is a summary of the outcomes:

-> Roof cladding materials alone (note metal roofing was always fitted with roof underlay (RU), as required by the NZBC, except where stated) :
-> All the long run materials were similar with 0.4 mm corrugate slightly better.
-> Painted and textured metal tiles were similar.
-> Concrete tiles (no RU) were good up to 1000 Hz but significantly worst at higher frequencies.
-> Concrete tiles (no RU) were worse than metal tiles (no RU) but when fitted with underlay both were
improved to a similar level (i.e. the underlay has a significant effect on reducing high frequency noise.
-> Plywood sarking on its own is similar to longrun but together the high frequency reduction is improved.
Roof cladding materials as above were installed over a 27° pitch roof cavity and a range of ceiling
-> The first thing to note is that all the installed roofs were significantly better than the roof cladding alone,
and as we half-expected, adjusting the ceiling construction is more important than changing the roof
-> The second thing is that all of the roof claddings have quite similar levels in each configuration
– which means that the roof system installation is always more important than the actual cladding.

There are graphs for the effect of standard GIB® against Noiseline® against two layers of GIB® and
standard bulk insulation against double the standard (note that the standard is much higher now than
it was when the original report was written in 2006), and the effect of plywood sarking directly under the

Change of pitch to 45°

The original intention was to repeat a series of tests using a different roof pitch. This would have altered
the reverberation in the roof cavity and changed the path for noise through the support members. A
start was made on this, and showed that there was a change in the frequency distribution attributed
to the different path, but this work was repeatedly interrupted by earthquakes and resources ran out.


-> The best effect and the cheapest method to improve noise reduction,through the roof/ceiling system, is to double the amount of bulk insulation (which of course has the collateral benefit of improving the heat insulation of the house – none of the other methods improves anything except noise.) This has the maximum benefit on the worst cladding, concrete tiles without underlay.

-> Then next is using plywood sarking under the roof cladding, which is probably the most expensive method and the one with other possible detriments. This is best for tiles with underlay.

-> Double skin of GIB® plasterboard (10 mm standard plus 13 mm standard) has a similar effect to
sarking except for the tiles.

-> Using 13 mm GIB® Noiseline plasterboard improves the tiles with underlay.

-> Changing the pitch of the roof may have an effect but this was not able to be fully investigated.


-> Different roof claddings do have different noise attenuation properties for traffic noise, as suggested in the 2006 report, but not in the order suggested by the report and generated by computer simulation.
-> Longrun roofing is best, presumably because of its airtight nature.

-> Concrete tiles with no underlay are worst, presumably because of the large air gaps.

-> Plywood does make an improvement but the most cost effective improvement with a desirable secondary benefit is the use of double bulk insulation. (Extra insulation is 30% of the cost of plywood sarking). All improvements to the roof/ceiling system are more important than the roof cladding itself.

Next steps: Field study

These studies do not include noise transmitted from the outside through the roof support members, or the
soffit. Although our roof structures did have rafters and purlins, they are not connected to the external