Measurements on Timber-Framed Floors with a Granular Material in the Floor Topping

th Proceedings of 20 International Congress on Acoustics, ICA 2010 23-27 August 2010, Sydney, Australia

Grant W. Emms (1)

(1) New Zealand Forest Research Institute, Private Bag 3020, Rotorua 3046, New Zealand. ([email protected])

PACS: 43.40.KD, 43.40.TM, 43.50.JH

ABSTRACT

Over the years timber-framed floor system designs have sometimes included some sort of granular material infill in order to reduce sound transmission between tenancies. Historically this has been in the form of some readily-available, low-cost material (e.g. ash, scoria, and sand). Recent research has been conducted into timber-framed floor toppings which contain a granular material infill in the form of a sand and sawdust mixture. The sand and sawdust mixture increases the mass of the floor, which improves the low- frequency impact insulation performance. This sand and sawdust infill also greatly increases the vibration damping in the upper part of the floor, improving the mid to high-frequency sound insulation performance, while also making the system robust to construction defects. This paper presents results of isolated element impact insulation and flanking transmission measurements of timber-framed floors which have a sand and sawdust mixture in the floor topping.

INTRODUCTION this granular material was and further communication with the authors did not reveal it, although they did say that it was This paper is an analysis of some recent measurements con- being installed in a block of apartments. However, they said ducted on timber-framed floor toppings (Emms et al, 2006; that the main reason for using this granular material was to Emms & Walther, 2010). The focus of this paper is the use of take advantage of the high damping afforded by the granular granular materials in floor topping system. The form of the material due to the friction between particles, because a granular materials considered is sand and sawdust mixtures. lightweight floor, in their opinion, could never have enough The granular material infill increases the mass of the floor, mass to provide excellent sound insulation. Figure 1 illus- which improves the low-frequency impact insulation per- trates this floor and shows how much granular filling there formance. The granular material infill also greatly increases was in the floor (i.e. lots). the vibration damping in the upper part of the floor, improving the mid to high-frequency sound insulation performance, while also making the system robust to construction defects. Floor walking surface GRANULAR MATERIALS IN FLOORS

Granular or particle-type materials have been used quite fre- quently in the past as an infill in timber structures. It was not uncommon to have sand or fly ash in timber floors in parts of Floor lower surface Granular filling the United Kingdom many years ago – the fill would be either placed on shelves between the joists or on the ceiling (which was made of material which could support the Figure 1. Swiss trial timber floor with upper floor surface weight). The primary function of this fill was for sound insu- floating on granular material (Walk and Keller, 2001). lation purposes. Another example of this is found in some old multi-storey timber buildings in New Zealand (e.g. the old Granular material vibration damping parliament buildings in Wellington), where volcanic scoria was used in the floors. More modern references to this are Since granular materials, particularly sand, have been used found in Switzerland, where sand has been used as a layer in with some success in buildings as a way of improving vibra- both retrofit and new buildings (Lappert & Geinoz, 1998). tion damping, it is worth overviewing some of the literature Incidentally, Lappert and Geinoz do note that the sand should which exists about this. It is clear from the literature that the be heated to ensure no living things are introduced. damping processes which occur in granular materials are complicated and not very well understood apart form some Walk and Keller (2001) did recent work to develop a floor specific instances. using a massive amount of ‘granular’ material on which a walking surface was floating. They do not say what exactly

ICA 2010 1 23-27 August 2010, Sydney, Australia Proceedings of 20th International Congress on Acoustics, ICA 2010 A good account of the use of sand with other mixtures of IMPACT INSULATION MEASUREMENTS ON materials has been provided by Kuhl and Kaiser (1962) for FLOORS WITH GRANULAR MATERIAL use with concrete structures. They tested various sands and INFILLS mixtures (fine sand, course sand, brick rubble, and mixtures of sand and sawdust), and found that hard granular materials In this section we describe measurements that were done on a with sharp edges or with a soft material (e.g. sawdust or rub- number of floor system elements which have a granular infill ber dust) gave better damping at lower frequencies. They in the floor topping. The granular infill used was either pav- thought that this was due to the sharp edges giving more ing sand, or a mixture of paving sand and sawdust. friction with vibrational strains and the lower wavespeed of vibrations in the sand sawdust or rubber dust mixtures. The The floor elements were tested in the same facility, with sup- sand/sawdust mixture tested was 80/20 sand/sawdust. They pressed flanking sound transmission, and in accordance with also note that the impedance of the granular fill should try to standard ISO 140-6. match the impedance of the surrounding, structural material to enable maximum coupling of energy into the granular fill. The following floors were tested:- They determined that the maximum damping was attained at cavity resonances in the granular fill. The damping was also • Basic Floor (Figure 13): A timber floor with ply- found to be non-linear in the sense that it depended on ampli- wood upper, solid timber floor joists, a fire-rated tude of the vibrations: higher amplitudes result in more ceiling disconnected from the joists using rubber movement of the granules and hence more friction. isolation clips, and filled with dense fibreglass.

Richards and Lenzi (1984) did measurements on the vibration • 40mm Sand Floor (Figure 14): The Basic Floor as a damping due to sand and found that below a strain of 10 -6 the subfloor with a floor topping consisting of 40mm particle movement is small and the damping in the sand is deep layer of sand between 45mm wooden battens, small and is probably not due to the friction of the particles; and another layer of plywood on the battens.The above that point damping becomes increasingly significant. sand density is 1250 kg/m 3. They also noted that loose sand in cavities is useful because there is the added damping benefit of having the granules • 40mm Sand/Sawdust Floor (Figure 15): The Basic move around from regions of high density to regions of lower Floor with a floor topping consisting of 40mm deep density. This movement is a maximum in resonances in the layer of sand and sawdust between 45mm wooden cavities which contain the granular fill. battens, and another layer of plywood on the battens. The sand-sawdust mix is 60% sand and 40% Xu et al. (2005) note that shear friction is the major contribut- sawdust by loose volume (density 1170 kg/m 3). ing factor to the damping due to granular fills. Since our con- cern in low-frequency floor vibrations is the bending waves • 85mm Sand/Sawdust Floor (Figure 16): The Basic in the floor which would cause a lot of shearing of any infill, Floor with a floor topping consisting of 85mm deep this is an important point to note. layer of sand and sawdust between 90mm wooden battens, and another layer of plywood on the bat- Sun et al. (1986) observed from experiment that, when look- tens. The sand-sawdust mix is 80% sand and 20% ing at sand laid on a vibrating metal plate, the vibrational sawdust by loose volume (density 1210 kg/m 3). damping in the plate is a maximum at frequencies above when the thickness of the sand is about equal to 0.05 λc, Measurement Results where λc is the longitudinal wavelength of the vibrations in In Figure 2 we can compare the results of the Basic Floor and the sand ( λc = c/f , where c is the propagation speed of the vibrations (100 to 200 m/s for sand) and f is the frequency). the 40mm Sand Floor. The greater mass and damping of the Below this point there is a sharp drop in damping, and above floor upper has resulted in significant improvements to the this point there is a gradual decrease (due to the increasing impact insulation performance. weight of the sand packing down the sand below and stop- It was noted before that the addition of sawdust with sand ping movement of the sand granules). Interestingly, sawdust does tend to improve the damping characteristics of sand is often mixed with sand when used for anechoic termina- alone. In Figure 3 we can compare the results of the 40mm tions for experiments with vibrations in beams. The sawdust Sand Floor and the 40mm Sand/Sawdust Floor. We note that is included to keep the sand from packing down, thereby although the mass has been slightly reduced by displacing improving its absorption characteristics. sand with sawdust, the extra damping has resulted in an im- Other vibration damping sources provement of performance, particularly at the higher frequencies. If vibration damping is important, another way to include this is to add a constrained viscous damping material, for example It was also noted that increasing the thickness of the like a viscoelastic polymer glued between plywood layers. sand/sawdust infill may improve vibration absorption at low This can be expensive, and the polymer’s damping perform- frequencies. In addition, increasing the sand/sawdust layer ance can be dependent on frequency and temperature, but it is will increase the mass, which will also improve the low- usually not dependent on vibration amplitude. frequency impact insulation performance. In Figure 4 we compare the results of the 40mm Sand/Sawdust Floor and the Yet another way to achieve more damping is to have layers 85mm Sand/Sawdust Floor. We see that the extra damping of material and to rely on the interaction between these layers and greater mass has resulted in an improvement of perform- to provide more damping, whether by friction between the ance for the mid to low frequencies. layers, air pumping, or fretting (where the surfaces become damaged due to rubbing, forming fragments, and thus con- It should be noted that the floor uppers with the granular tributing to frictional effects). infill are not floating on the subfloor in any way. The separation of the additional layer of plywood is achieved through battens which are directly fixed to the subfloor surface. And the top layer of plywood is directly fixed to the battens. There is no resilient isolation between layers. All the isolation is

2 ICA 2010 23-27 August 2010, Sydney, Australia Proceedings of 20th International Congress on Acoustics, ICA 2010 achieved through vibration absorption in the granular mate- FORCED VIBRATION MEASUREMENTS rial. It is an inherently robust system. In the previous section it was stated that the granular infills Basic Floor; Ln,w (Ci ) = 61(-1) dB 40mm Sand; Ln,w (Ci ) = 52(0) dB are able to absorb the vibrations generated in the floor (presumably by impacts). This can be shown experimentally from 80

, d B surface vibration measurements.

n 75 L 70 65 On the floor an electrodynamic shaker was used to provide a 60 vertical force on the floor upper surface. The shaker was 55 connected to the floor through a wire stinger and a reference 50 45 force transducer. A scanning laser vibrometer (Polytec PSV 40 300) was used to measure the velocity of the floor and ceiling 35 normal to the surface. A grid with a spatial resolution of 10- 30 14cm was used to obtain a map of the surface velocity of the 25 floor and ceiling relative to the input force; both amplitude 20 15 and phase information was recorded at each frequency.

Normalized10 impact sound pressure level, Figure 5 shows the results of the vibration measurements on 50 63 80 100 125 160 200 250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 the ‘85mm Sand/Sawdust Floor’. We can see how the vibra- Third octave band (Hz) tion is quickly damped by the granular infill. Figure 2. Comparison of the impact insulation perform- In comparison, Figure 6 shows the results of the vibration ance of the ‘Basic Floor’ and the ‘40mm Sand Floor’. measurements on a floor with a floating gypsum concrete Both floors are 3.2m by 7m (joist length). screed (Figure 17). We can how the vibration is not damped, and spreads over the whole floor. Carefull edge detailing is 40mm Sand/sawdust; Ln,w(Ci ) = 52(-1) dB 40mm Sand; Ln,w(Ci ) = 52(0) dB therefore required to avoid severly reducing the floors sound

80 insulation performance. 75 , d B n

L 70 Displacement per unit force at 190Hz, and at phase 90° relative to force 65

60 -7 x 10 55 50 45 1.5 40 35 1 30 0.5 25 0 20 -0.5 15 Normalized impact sound pressure level, -1

10 Vertical displacement (m) -1.5 0 5 63 80 25 60 00 50 15 30 00 0 0 0 100 1 1 2 2 3 400 500 6 8 0 5 000 0 10 12 1600 2000 2500 3150 4 50 Third octave band (Hz) -1 -2 Figure 3. Comparison of the impact insulation perform- -3 -1 -4 -2 ance of the ‘40mm Sand/Sawdust Floor’ and the ‘40mm -5 -3 Sand Floor’. Both floors are 3.2m by 7m (joist length). y (m) x (m)

40mm Sand/sawdust; Ln,w(Ci ) = 51(-1) dB 85mm Sand/sawdust; Ln,w(Ci ) = 48(-2) dB Figure 5. Measured vibration (displacment) of the 85mm 80 Sand/ Sawdust floor upper surface and ceiling from 1N of 75 , d B

n forced vibration on the upper surface.

L 70 65 60 Displacement per unit force at 190Hz, and at phase 0° relative to force

-8 55 x 10 50 45 6 40 35 4 30 2 25 20 0 15 -2 Normalized impact sound pressure level, 10

Vertical displacement (m) -4 0 5 63 80 25 15 00 0 0 0 100 1 160 200 250 3 400 500 630 8 000 00 1000 1250 160 2 2500 3150 400 5 -6 Frequency, f , Hz -1 -2 Figure 4. Comparison of the impact insulation perform- -3 -1 ance of the ‘40mm Sand/Sawdust Floor’ and the ‘85mm -4 -2 -5 Sand/Sawdust Floor’. Both floors are 3.2m by 5.5m (joist -3 y (m) x (m) length). Figure 6. Measured vibration (displacment) of a Floating Gypsum Concrete floor upper surface and ceiling from 1N of forced vibration on the upper surface.

ICA 2010 3 23-27 August 2010, Sydney, Australia Proceedings of 20th International Congress on Acoustics, ICA 2010

Measurement results

FLOOR/WALL SYSTEM MEASUREMENTS In the following figures we compare the sound insulation measurement results of the two floor toppings (i.e. 40mm The performance of a floor/wall system with a granular mate- sand/sawdust and 20mm particleboard). rial-infilled floor topping was measured in the BRANZ four- room flanking test facility. The flanking facility was designed We can see that the sand/sawdust mixture markedly improves to test the sound insulation of a horizontal cruciform the horizontal insulation performance (Figure 9 and Figure wall/floor system (Figure 7). All other sound paths were sup- 11). Presumably this is due to a significant reduction in vibra- pressed by the test facility (Emms & Walther, 2010). The tion energy being transmitted to the subfloor and across the measurements were done in accordance with standard ISO continuous floor diaphragm. The horizontal airborne insula- 140. tion performance of the system (Figure 11) with the 40mm sand/sawdust topping system is limited by the wall performance; whereas the sound transmitted via the floor flanking path limits performance in the 20mm particleboard system.

The sand/sawdust mixture also improves the vertical insula- Wall Linings Floor Topping 2x13mm Fire-rated tion performance (Figure 10 and Figure 12). This improve- Plasterboard. ment is due to a combination of increased mass and damping in the floor topping reducing energy being transmitted to the ceiling and to the frame and wall.

Horizontal Impact Insulation

40mm Sand/Sawdust Floor (LnT,w= 43) 20mm Particleboard Floor (LnT,w= 55)

60 Figure 7. The form of the specimen tested for flanking 50

transmission. 40 LnT(dB) The system under test used the floor’s 17mm plywood as a 30 structural floor diaphragm. Hence there was significant acoustic bridging for horizontal sound transmission. The wall 20 consisted of a double-stud construction, with two layers of 10 0 3 0 0 0 0 0 5 6 8 00 60 00 15 0 30 0 50 00 50 1 125 1 2 250 3 4 500 6 8 000 000 13mm fire-rated plasterboard on each side (Figure 7). The 1000 12 16 2 250 31 4 500 subfloor construction is shown in Figure 8. On the subfloor a Third octave band (Hz) number of floor topping systems were trialled. The measurement results of two floor toppings are examined:- Figure 9. Horizontal impact sound insulation measurements. • 20mm particleboard screwed to the subfloor. Vertical Impact Insulation 45mm deep battens screwed to subfloor, 40mm of • 40mm Sand/Sawdust Floor (LnT,w=50) 20mm Particleboard Floor (LnT,w=60) sand/sawdust (80% / 20% by loose volume) infill 80 between battens, and 20mm particleboard screwed to top of battens. Edges are constructed by butting 70 battens against walls –no foam separation was 60

used. 50

LnT(dB) 40

20 1 layer 17 mm plywood ( Floor Upper ) 10

0 3 0 0 0 0 0 5 6 8 00 60 00 15 0 30 0 50 00 50 1 125 1 2 250 3 4 500 6 8 000 000 1000 12 16 2 250 31 4 500 Third octave band (Hz)

75 mm fibreglass infill 245mm x 45mm (R1.8) ( Cavity i nfill ) Timber I-beam Joists Figure 10. Vertical impact sound insulation measurements.

RSIC-1 Resilient Clips

Steel furring cha nnels. 2 x 13 mm plasterboard (25kg/m 2). (Ceiling )

Figure 8. The basic subfloor used for the flanking transmission measurements.

4 ICA 2010 23-27 August 2010, Sydney, Australia Proceedings of 20th International Congress on Acoustics, ICA 2010

Horizontal Airborne Insulation Thanks to Tony Walther and Graeme Beattie for the opportu- nity to make the flanking transmission measurements at 40mm Sand/Sawdust Floor (DnT,w=59) 20mm Particleboard Floor (DnT,w=54) BRANZ, Judgeford, New Zealand. 80

60 REFERENCES 50 Emms, G., Chung, H., Dodd, G., McGunnigle, K., Schimd, 40 DnT(dB) G., (2006) Maximising impact sound resistance of timber

30 framed floor/ceiling systems, Forest and Wood Products Research and Development Corporation (Australia), Pro- 20 ject PN04.2005. 10 Emms, G. W.; Walther, T.; (2010). Flanking Transmission 0 3 0 0 0 0 0 5 6 8 00 60 00 15 0 30 0 50 00 50 1 125 1 2 250 3 4 500 6 8 000 000 1000 12 16 2 250 31 4 500 Measurements: Part 1- Measurement Techniques. Build- Third octave band (Hz) ing Acoustics , Volume 17, Number 1, Pages 15–34 Kuhl, W., Kaiser, H. (1962). “Absorption of structure-borne Figure 11. Horizontal airborne sound insulation measure- sound in building materials without and with sand-filled ments. cavities”, Acustica , 2, 179-188. Lappert, A., Geinoz, D., (1998). “Experience in Multi-storey Vertical Airborne Insulation timber buildings consulting and field measurement”, 40mm Sand/Sawdust Floor (DnT,w=62) 20mm Particleboard Floor (DnT,w=58) Proceedings of Acoustic Performance of Medium-rise 100 Timber Buildings , Dublin, Ireland, Dec. 1998. 90 Richards, E. J., Lenzi, A. (1983). “On the prediction of im- 80 pact noise, VII: The structural damping of machinery”, 70 Journal of Sound and Vibration , 97(4), 549-586.

60 Sun, J.C., Sun, H.B., Chow, L.C., Richards, E.J. (1986).

50 “Predictions of total loss factors of structures, part II: DnT(dB)

40 Loss factors of sand-filled structure”, Journal of Sound

30 and Vibration , 104(2), 243-257.

20 Walk, M. & Keller, B. (2001). “Highly sound –insulating

10 wooden floor system with granular filling”, Proceedings

3 0 5 0 0 0 50 6 80 00 0 50 1 00 00 00 50 1 125 160 2 2 3 4 5 630 8 500 000 of ICA 2001. 1000 12 1600 200 2 315 4 500 Third octave band (Hz) Xu, Z., Wang, M. Y., & Chen, T. (2005). Particle damping for passive vibration suppression: Numerical modelling Figure 12. Vertical airborne sound insulation measure- and experimental investigation. Journal of Sound and Vi- ments. bration , 279(3-5), 1097-1120.

CONCLUSION

In this paper the use of granular materials (specifically sand and sand/sawdust mixes) in floor topping systems have been considered. Recent measurement results were compiled and examined to show that these sorts of systems can signifi- cantly improve the sound insulation performance of the floor. The improved performance is due to increased vibration damping and due to increased mass. As a result of these two factors, performance is improved across the whole frequency range.

It was also shown that the vibration damping effect localises impact vibrations, thereby reducing flanking transmission issues.

Another advantage such floor topping systems with granular infill is that they are not floating on the subfloor in any way. The topping is directly fixed to the subfloor, and there is no resilient isolation between layers. Edges are simply constructed by butting battens against walls – again no foam separation is used. All the isolation is achieved through vibration absorption in the granular material. It would appear, therefore, that it is an inherently robust system.

ACKNOWLEDGEMENTS

Thanks to George Dodd, Ken McGunnigle and Gian Schmid for helping to make the floor element measurements at the University of Auckland.

ICA 2010 5 23-27 August 2010, Sydney, Australia Proceedings of 20th International Congress on Acoustics, ICA 2010 APPENDIX: FLOOR DIAGRAMS

Figure 16. The ‘85mm Sand/Sawdust Floor’.

Figure 13. The 'Basic Floor' on which toppings were added.

Figure 17. Floating Gypsum Concrete Floor.

Figure 14. The ‘40mm Sand Floor’.

Figure 15. The ‘40mm Sand/Sawdust Floor’.

6 ICA 2010