Proceedings of the Institute of Acoustics

MAKING SENSE OF THE ACOUSTIC PERFORMANCE OF

FLANKING PATHS FOR CURTAIN WALLING SYSTEMS

Nick Conlan Apex Acoustics Ltd, Gateshead, UK

Jack Harvie-Clark Apex Acoustics Ltd, Gateshead, UK

1 ABSTRACT
The sound insulation between spaces can often be limited by the flanking transmission of the
curtain wall elements of the façade. Glazed curtain walling is used for many commercial and
residential projects; it is important to understand the limitations to sound insulation that the curtain
walling introduces.

This paper looks at how the different types of performance data provided for curtain walling systems
are used to predict the sound insulation between spaces. It shows how the sound reduction index
and the flanking level differences can both be used for calculations, and explains how the two
parameters are linked.

Common types of curtain walling systems and the associated flanking paths are described, along
with typical performance achieved and methods for improvement. A case study is included to
demonstrate how site measurements can be used to establish the contribution from flanking paths.

The paper also demonstrates how the introduction of glazing to the curtain wall sections can
change the measured sound reduction of the sections, and why Rw values of curtain walling
element should be used with caution when estimating the flanking of curtain walling systems.

2 TYPES OF CURTAIN WALL SYSTEMS

Curtain walling is a non-structural system that weatherproofs the outside of buildings and supports
no load other than its own weight and the environmental forces which act upon it. The framework
has vertical elements known as mullions and horizontal sections known as transoms. Glazing
panels are fitted within the framework, but they can also be fitted with louver openings for ventilation
or opaque spandrel panels.

There are two basic types of curtain walling installations: stick and unitised.

2.1 Stick Systems

Stick systems are constructed on site. They get their name from the fact that the vertical structural
mullions (sticks) are fixed first. After the mullions are secure, the horizontal transoms are added and
then the glazing panels, spandrels and vents installed in the completed grid.

2.2 Unitised Systems

Unitised curtain walling systems are installed as a series of factory-assembled frames, usually with
interlocking mullions and transoms. The glazing panels and spandrels panels are pre-fitted and
seals are also applied or prepared in the factory.

Unitised systems are popular because they are quicker to install on site and they eliminate, or
reduce, the need for on-site sealing, therefore making them less reliant on the standard of site
workmanship.

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

3 FLANKING PATHS
The flanking paths for sound to pass from one space to another can be considered to be any route
other than through the separating element. For most scenarios this can be broken down to be the
sound through the glazing, through the curtain wall sections (mullions or transoms) and the method
of sealing between the curtain wall sections and the main separating element (floor or partition).
These are shown in Figure 1.

Figure 1: Plan view of flanking and direct paths for sound transmission

When considering vertical sound insulation through a floor, there is also the additional path of sound
flanking through common mullions, which are not split at the separating floors. This is shown in
Figure 2.

Figure 2: Section showing the flanking path via a continuous mullion

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

3.1 Performance of curtain wall hollow sections (mullion or transom)

There are three types of curtain wall hollow sections considered in this paper and they are shown in
Table 1.
Table 1: Typical details of curtain wall hollow sections
Stick systems Unitised systems
One-piece hollow section Split hollow section Independent frames

Curtain walling suppliers will have their own profiles and it is possible for systems with the same
external dimensions and to have different internal profiles. The wall thickness of the section can
also differ between suppliers and the method of sealing between the unitised systems can vary.

3.1.1 Single piece hollow sections

Typical Dn,f,w values for hollow sections between 110mm and 200mm deep are provided in Table 2,
along with the R’w measured for one section. The performance of single piece hollow sections are
often influenced by a dip in the level differences at a resonance frequency that changes with the
depth of the section, and these frequencies are shown where they could be established from the
test data. The superscripts in the table refer to the references from which the test data is derived.

Table 2: Acoustic performance of single piece hollow sections

Acoustic Frequency of sound
Section details
performance / dB insulation dip / Hz
125mm transom1 38 Dn,f,w 630
125mm mullion1 44 Dn,f,w 1000
200mm mullion1 40 Dn,f,w 315
110mm mullion2 42 Dn,f,w -
190mm mullion2 40 Dn,f,w -
110mm mullion3 44 Dn,f,w 1600
120mm mullion4 41 R’w 1000

There is a noticeable difference in the D n,f,w values between the 125mm transom and 125mm
mullion. Although the external dimensions of the hollow sections are the same and the internal
profiles look similar, but not identical, it is not known if they have different wall thickness’. When the
transom was tested there were three mullions flanking the separating element and size of the
glazed panels either side of the hollow section were larger for the testing of the mullion.

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

3.1.2 Split hollow sections

The sound insulation provided by typical split hollow sections are shown in Table 3. The split hollow
sections don’t have the same distinctive dip in performance at a resonant frequency, yet they don’t
perform significantly better overall when compared to the one piece hollow sections.

Table 3: Acoustic performance of split hollow sections

Section details Acoustic performance / dB
125 mullion1 47 Dn,f,w
110 mullion2 44 Dn,f,w
170 mullion5 36 STC

3.1.3 Independent frames

The independent unitised systems have greater isolation between the aluminium frames as they
can be independently supported and only connected via flexible gaskets. The typical sound
insulation from independent framed systems is shown in Table 4.

Although the split mullion section with a gap is not a realistic system, it does demonstrate a
difference of 11dB between a split mullion system without a gap and a system with the same frames
supported independently with a 25mm gap between the frames.

Table 4: Acoustic performance of independent frames

Section details Acoustic performance / dB
125 section6 50 Dn,f,w
160 mullion2 54 Dn,f,w
170 split mullion with 25mm gap5 47 STC

3.2 Performance of glazing

The interaction between the glass and the curtain wall hollow section is important. The dynamics of
the glass and section are coupled; sound incident on the glass displaces as a membrane which
applies bending at the boundary condition (mullion) and excites the membrane on the opposite
side.5

In situations where the flanking via other routes is controlled to lower levels the change in
performance for different glazing types can be seen. The inner pane has the greater influence in the
performance and the higher the Rw value of the glass the better the control of flanking. The Dn,f,w
values for three glazing configurations are shown in Table 5.

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

Table 5: Acoustic performance of the flanking via glazed units

Section details Glazing Acoustic performance dB
125mm overclad single piece hollow section1 8-16-4 52 Dn,f,w
125mm overclad single piece hollow section1 4-16-8 54 Dn,f,w
125mm overclad single piece hollow section1 6-12-9GH 59 Dn,f,w
190mm overclad single piece hollow section2 6-16-9GH 59 Dn,f,w
Another study looked at how the glazing affected the performance of the mullion. The STC of the
mullion was measured without any glazing, then it was retested with 9-25-12 glazing. The measured
levels are shown in Table 6.

Table 6: Change in the sound insulation of a mullion with glazing installed

STC without glazing STC with glazing Dn,f,w with glazing
Treatment to 170mm mullion
/ dB / dB / dB
Filled and overclad 52 42 58
Filled 38 37 53
No treatment 36 32 48

These measurements demonstrate that the flanking via the glazing can be as high as 59 dB Dn,f,w
when flanking via the hollow section is suitably controlled. It does however also show that the sound
reduction of the hollow sections, treated or untreated, is reduced when the sections are tested with
glazing in situ, as opposed to testing the section independently without glazing. This suggests that
the flanking path via the glazing and the hollow section, Figure 3, is important for determining the
overall sound insulation of the system.

Figure 3: Flanking path via glazing and hollow section interaction

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

3.3 Performance of mullions for vertical transmission

Sound can travel between floors via a mullion which is exposed in the rooms either side of the
separating floor. The mullions can be continuous structures, or can be split and supported at the
floor edge, and the split mullions can have infill pieces inserted to blank off airborne sound
transmission. Typical performance for three 2.75m long,125mm deep mullions are shown in Table
7.

Table 7: Performance of mullions for vertical transmission

Mullion details Acoustic performance / dB
Continuous and unbroken 43 Dn,f,w
Broken at floor level and joined with Schuco profile infill 54 Dn,f,w
Broken, partially filled with foam and ends capped 68 Dn,f,w

The case study in Section 6 is used to estimate the level of sound flanking via continuous mullions.
For two continuous mullions, 100mm deep and 2.5m long, the predicted D n,f,w value is 51 dB. In the
case study each mullion is partially enclosed into the façade and this is likely to provide some
damping to the mullions.

3.4 Performance of the junction from the main partition to the curtain wall
It should be practical for a partition to be sealed to the curtain walling section and not reduce the
performance of the section. Gaps up to 25 mm can be sealed with backer rods and a flexible
sealant, without reducing the measured sound reduction of the tested sections5. Care should be
taken to keep the gaps to a minimum and for the sealant to be non-setting as the curtain wall is
designed to move and sealants will crack if they are set.

For scenarios where a slimmer wall section is preferred close to the window, a reduced
performance would be expected and some proprietary products are marketed for this purpose, such
as Siderise FIP high performance acoustic panel which is 31 mm thick and has a sound insulation
performance of Rw 46 (-2;-3) dB7.

Various staggered plate systems have been tested5 which allow for movement of the curtain wall
sections in two axis. 15mm plasterboard and 3mm aluminum plate systems have been tested to
achieve Rw 51 dB and Rw 49 dB respectively.

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

4 PREDICTING ON-SITE PERFORMANCE FROM TEST DATA

4.1 Comparing R and Dn,f values
The performance of a curtain walling system may be presented as a D n,f,w value, as defined in BS
EN ISO 10848-28 and is tested in a laboratory. The sound reduction R, is also established in a
laboratory, with a curtain wall section of known area installed into a high performing wall which
performs significantly better than the curtain wall section.

If we consider a mullion being tested in a laboratory, and the mullion has an area S, within the
larger test wall, then:

Rmullion = L1 – L2 + 10 log Smullion / A, where S is the area of the mullion (m2)

The relationship for flanking sound between two rooms is according to ISO 10848-2:

Dn,f = Df - 10 log A / A0 where A0 is 10m2 and Df = L1 - L2

Combining the two equations gives:

Rmullion - 10 log Smullion / A = Dn,f + 10 log A / A0

Rmullion = Dn,f + 10 log A/A0 + 10 log Smullion/A
Rmullion = Dn,f + 10 log (Smullion.A / A.A0)
Rmullion = Dn,f + 10 log (Smullion/10)

4.2 Predicting the level difference between rooms

The on-site performance can be considered using the Apex Method9 of predicting the contribution
from each of the routes to find the overall performance. So if we can establish the partial apparent
level difference due to each of the elements we can calculate the global level difference. Equally, as
demonstrated in section 6, once we have measured the level difference between two rooms, we
can estimate the partial performance of an element

The partial Dn,T,w values can be calculated using the following equations

Dn,T,w = R’w + 10 log (V/S) -5

Dn,T,w = Dn,f,w +10Log(V)-15

When using Dn,f,w values for predictions, the length of tested construction needs to be known. If the
length of the predicted section is different then the Dn,f,w value used in the calculation should be
adjusted to reflect the difference in lengths. For example, if a tested transom was 2.7 m long in the
laboratory, but for a prediction the transom formed 10.8 m of the perimeter of a room, the Dn,f,w
value of the transom would be reduced by 6 dB.

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

5 METHODS TO IMPROVE PERFORMANCE

5.1 Internal treatments
Internal treatments can consist of specific products or bespoke build ups, which can introduce
further mass to the construction, absorption, damping or a combination of these. Test data is
available which provides Rw and STC values for the aluminum section however these must be
adjusted to allow for the effect of the glazing when considering the overall flanking performance.
Typical treatments and the level of improvement are shown in Table 8.

Table 8: Improvement provided by internal treatments to hollow sections

Treatment Untreated / dB Treated / dB Improvement
200mm hollow section filled with 2 layers of
40 Dn,f,w 56 Dn,f,w 16 dB
fermacell and compressible infill1
170mm split hollow section with mineral wool
36 STC 38 STC 2 dB
and mass barrier material5
125mm section filled with mineral wool and
44 Dn,f,w 47 Dn,f,w 3 dB
4kg/m2 damping material3

200mm section with Siderise HB insert -MI610 33 Rw 41 Rw 8 dB

5.2 External treatments

External treatments consist of materials over cladding the aluminum sections. These can consist of
elements providing additional mass, absorption and damping and typical levels of improvement for
different systems are shown in Table 9.

Table 9: Improvement provided by external treatment to hollow sections

Treatment Untreated / dB Treated / dB Improvement
125mm mullion with 20mm timber and 15mm
44 Dn,f,w 50 Dn,f,w 6 dB
mineral wool to one side1
125mm mullion 20mm timber and 15mm
44 Dn,f,w 52 Dn,f,w 8 dB
mineral wool to both sides1
3mm aluminum plate and 5mm vinyl mass
36 STC 48 STC 12 dB
barrier 10kg/m3 both sides5
15mm plasterboard and 5mm vinyl mass
36 STC 50 STC 14 dB
barrier both sides5

15mm plasterboard both sides fixed direct5 36 STC 42 STC 6 dB

110mm section 2mm steel, 4kg/m2 damping

44 Dn,f,w 50 Dn,f,w 6 dB
material3
5.3 Additional hollow sections
When a single section is used, the performance will be limited by the glazing, no matter how well
the sections are treated. Where higher levels of sound insulation are required a second hollow
section must be introduced to reduce the flanking through the glazing. Dn,f,w values of 67 / 68 dB can
be achieved1,6 with both sections overclad; and for completed rooms NIC values of 62 to 68 dB
have been achieved11.

As the flanking performance is increased the sound transfer via the separating element or the
method of sealing the curtain walling also become more important and some test data may
underestimate the flanking performance for these reasons.
Vol. 38. Pt. 1. 2016
Proceedings of the Institute of Acoustics

6 CASE STUDY - VERTICAL TRANSMISSION VIA A MULLION

AND TRANSOMS
This case study was for a residential development which included curtain walling elements within
the façade build up. A section through the façade and a photograph of the external elevation is
shown in Figure 4.

Figure 4: Section through the floor and façade junction and curtain walling photograph

Two mullions run continuously between floors. The non-curtain wall elements have a 140 mm
internal block lining and there are two transoms within the floor build up. During the testing it was
clear that the sound was flanking via the window area. The curtain wall system was 100mm deep
and there was a dip in the sound insulation at 2500 Hz. Two rooms of the same construction were
tested on site and they achieved DnT,w 52 (-1;-3) dB and DnT,w 51 (-1;-5) dB for the floor airborne
tests.

The Dn,f,w of the exposed transoms can be taken from test data1 and it is estimated to be Dn,f,w 53 dB
for the lengths of transoms exposed. The room volumes were approximately 50 m3 and the
separating floor area was 20 m2.

If we use the equations shown in section 4.2 we can estimate the contribution of the flanking via the
mullions as we know the overall D nT,w and can estimate the contribution for the floor element and
the exposed transoms. The resulting Dn,f,w is shown in Table 10. This method clearly highlights the
contribution from each sound transmission path to the overall level difference achieved.

Table 10: Calculation of Dn,f,w from tested DnT,w values

Element Area /m2 R’w or Dn,f,w /dB Partial DnT,w Global DnT,w

Mullions - 51 53

Transoms - 53 55 51

Floor 20 65 64

Vol. 38. Pt. 1. 2016

Proceedings of the Institute of Acoustics

7 CONCLUSIONS
The sound insulation between spaces can often be limited by the flanking transmission of the
curtain wall elements of the façade. If the Dn,f,w values for the system are known, the on-site
performance can be predicted, however test data is often not available until suppliers are confirmed.
Initial estimates of the flanking performance can be made using typical performance figures, but
these must be used with caution as there can be significant differences in performance between two
curtain wall sections which have the same dimensions. It is vital to appreciate the details of the
proposed system and use suitable test appropriately in order to reliably predict the potential sound
insulation performance.

When Rw values are used for the performance of a curtain wall hollow section, the predictions
should make allowance for a reduced performance once the glazing is installed in the system as the
flanking via the junction of the glazing and the curtain walling hollow sections needs to be
considered.

.
8 REFERENCES
1. Summary of results of research project ‘Insulation against noise transmission for façades’
Schuco International, Laing, 2000
2. Wictec Technal 0151 Design manual - Acoustics
3. Improvement of acoustic flanking transmission through lightweight facades Louwers, Marc
Internoise 2012
4. Analysis of direct and flanking sound transmission between rooms with curtain wall facades.
6th international building physics conference IBPC 2015, Zuccherini, Martello, Secchi,
Fausti, Cellai, Santoni
5. Lateral sound flanking transmission at curtain wall mullions: an empirical investigation to
identify controlling mechanisms, Elizabeth Valmont 2015 Phd
6. Horizontal and vertical flanking transmission and sound transmission loss -Schuco Skyline
S 65F Unitised construction, August 2004
7. Siderise FIP panel data sheet,
http://www.siderise.com/applications/products/fip-high-performance-acoustic-panel?interiors
8. BS EN ISO 10848-2:2001 Acoustics - Laboratory measurement of the flanking transmission
of airborne and impact noise between adjoining rooms- Part 2: Application to light elements
when the junction has a small influence
9. Practical Acoustic Design – The Apex method. Proceedings of the Institute of Acoustics, Vol
36.Pt 3 2014 J Harvie-Clark, available here.
10 Siderise MI6 HB Inserts data sheet
http://www.siderise.com/applications/products/mullion--transom-acoustic-inserts?facades
11 The role of acoustics in curtain wall design, strategies to achieve desired performance
criteria, Quebec building envelope council, annual conference april 2015, John LoVerde,
VeneklasenAssociates http://www.cebq.org/documents/RoleofacousticsinCWdesign.pdf

Vol. 38. Pt. 1. 2016