Sustainable acoustic absorbers from the biomass

Applied Acoustics 72 (2011) 350–363

Contents lists available at ScienceDirect

Applied Acoustics journal homepage: www.elsevier.com/locate/apacoust

Sustainable acoustic absorbers from the biomass David J. Oldham ⇑, Christopher A. Egan, Richard D. Cookson Acoustics Research Unit, School of Architecture, University of Liverpool, Liverpool L69 3BX, UK

a r t i c l e

i n f o

Article history: Received 23 March 2010 Received in revised form 20 December 2010 Accepted 21 December 2010 Available online 4 February 2011 Keywords: Sound absorbers Sustainable Biomass Natural fibres Reeds

a b s t r a c t There is currently considerable interest in developing sustainable absorbers, either from biomass materials or recycled materials, and it is the former that is the subject of this paper. A number of potential candidate materials are available from the biomass in the form of organic fibres. Non-fibrous materials, such as configurations of whole straw or reed, can also act as sound absorbers. A combination of impedance tube and reverberation chamber measurements have been carried out for a number of biomass materials and the effectiveness of current models for the prediction of the absorptive properties of natural fibres has been investigated. Examination of the acoustical characteristics of a range of natural fibres has confirmed their effectiveness as porous sound absorbers and also the limitations of current models for predicting their performance. Examination of the acoustical performance of materials consisting of different configurations of whole reeds and straws has revealed that these also possess considerable potential for application as broadband sound absorbers with particularly good low frequency absorption characteristics. The combination of natural fibres and whole reeds offer the possibility of developing a range of sustainable absorbers which act very effectively across the complete audio frequency range. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Acoustic absorbers are widely used in noise control to reduce sound reflected from surfaces. This paper describes work carried out as part of the EU sponsored Holiwood project to identify a sustainable absorptive treatment to complement the sustainable structural components of a proposed noise barrier system made from thermally treated European hardwood. There is currently considerable interest in developing sustainable absorbers, either from biomass materials [1,2] or recycled materials such as crumb rubber [3–5]. In this paper the properties and potential of the former are examined. The term biomass refers to living and recently dead biological material. Biomass materials are inherently sustainable as they constitute part of the normal carbon and nitrogen cycles. Thus, provided only renewable energy is employed during processing, they are virtually carbon neutral. Similarly, if the use of highly toxic chemicals for protection against deterioration due to biological attack from fungi or insects is avoided then they can be disposed of after use by returning to the natural cycles by simple composting. In investigating the potential of biomass materials as sound absorbers, the first step was to identify the characteristics of conventional absorbers from an examination of earlier work. A number approaches are currently used to obtain a sound absorbing finish. One approach is by the use porous materials, frequently ⇑ Corresponding author. Tel.: +44 151 794 7576; fax: +44 151 794 2605. E-mail address: [email protected] (D.J. Oldham). 0003-682X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.apacoust.2010.12.009

consisting of glass fibre or mineral wool in which some form of binding agent, known as a binder, is used to hold the fibres together and to maintain dimensional stability. Another approach to ensuring structural integrity and dimensional stability is to produce rigid panels constructed with open textured porous materials. There are a number of potential candidate materials available from the biomass in the form of organic fibres, either vegetable or animal in origin, and these constitute obvious candidates for the direct replacement of the glass and mineral wool fibres used for conventional sound absorbers. Many of these have a long history of use in fabrics (cotton, wool, flax, silk), as floor coverings (wool, reeds), sacking (jute, hessian) and ropes (hemp) [6]. The sound absorption characteristics of a selection of these materials will be examined in this paper. A number of models are available for predicting the acoustical characteristics of fibrous materials, the most well known being that derived by Delany and Bazely [7] which is based upon measurements carried out on glass and mineral wool fibres. However, of particular interest in the context of this work is a more recent model derived by Garai and Pompoli [8] which is based upon measurements made on polymer fibres whose diameters and density of the matrix material are both closer to those of natural fibres. Hence the applicability of both empirical models for predicting the absorption coefficients of fibrous absorbers obtained from the biomass is also investigated. Layers of un-shredded straw or reeds are also porous in nature with pores arising from gaps between the individual stems and the non-uniformity of the cross sections of individual stems. This

351

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

results in a large number of slit like pores, parallel to the straw or reed lengths, which communicate with large cavities between the individual straws or reeds. The potential of reeds or straw to act as efficient sound absorbers was of particular interest in this work because these materials have been shown to be durable when subjected to extreme environmental conditions similar to those likely to be experienced by a highway noise barrier [9].

2. Biomass options for sustainable noise absorbers 2.1. Fibrous materials There are many potential candidate materials in the form of organic fibres, either vegetable or animal in origin, and information regarding a number of these, most of which have been investigated in this work, are shown in Table 1. Plant fibres are long, thick-walled cells whose composition is principally cellulose. The cellulose molecules consist of glucose units linked together in long chains, which in turn are linked together in bundles called micro-fibrils, glued together with pectin. Hemicelluloses and lignin are also found in all plant fibres. Hemicelluloses are polysaccharides bonded together in relatively short, branching chains. They are intimately associated with the cellulose micro-fibrils, embedding the cellulose in a matrix. Lignin is the compound which gives rigidity to a plant and is responsible for the heights and the rigidity found in crops such as straw or reeds. Plant sources of fibre include cotton, hemp, kenaf, ramie, sisal, flax, linen, lime, jute, sea grass, bamboo and abaca. In general, natural fibres can be grouped into two categories: soft fibres and hard fibres. Most soft fibres come from the bast portion of the plant which lies directly under the outer bark or skin and is where the transport of the products of photosynthesis and the development of stabilizing structures take place. Hemp, flax, jute and ramie are soft fibres. The fibres are usually freed from the stalk by a process termed retting. The released fibre bundles are frequently used without additional separation, however, flax and ramie strands, are usually separated into individual fibre cells, or true plant fibres. Hard fibres are comprised not only of the bast but also partly of the hardened wood core of the plant and the hardness of the fibres is caused by the deposit of lignin in the cell walls. Hard fibres generally come from the leaves of monocot (single seed-leaf) species, for example sisal agave, fibre banana and palms. There is a great variation in fibre dimensions and the various types of raw material are frequently separated using processes to generate fibres suitable for specific end products, e.g., bast or stem fibres are mainly used in the textile or rope industries because both require long fibres. From examination of Table 1 it can be seen that the basic plant fibres can be divided into two sub groups depending upon the

mean fibre diameter. Jute, hemp and sisal and are relatively coarse, with fibre diameters approaching 100 lm for the jute and hemp, and with a value in excess of 200 lm for the sisal fibres. The cotton, flax and ramie have fibre diameters in the range 13–25 lm. It is the finer nature of these fibres that has resulted in their use for clothing. The density of the matrix material in all cases, however, only varies slightly which reflects the fact that they are all mostly composed of cellulose. The bulk density varies between the samples and, given the relatively slight variation in the densities of the matrix material, this reflects the degree of compaction to which the samples have been subjected and possibly the use of a significant amount of binder. Thus, the hemp batts, a product manufactured as animal bedding, were bound using a high proportion of latex which was clearly visible on micrographs. The binding process appears to have been performed under great pressure to give a material with a relatively high bulk density compared to the hemp batts where approximately 15% constituted recycled polyester which acted as a binder. In contrast the bulk density of the sisal, as used for the acoustic measurements described in this paper, was low as it was hand compacted and, given the coarseness and stiffness of the fibres, only a relatively limited degree of compaction could be achieved. Animal sources of fibre include sheep, alpaca, llama, goat, and camel, and can be either wool or hair. Insect fibre is predominantly from silkworm cocoons. There is considerable interest in finding new applications for wool as developments in synthetic fibres have adversely affected its original market as a fibre for clothing to the extent that wool is now a by-product of the farming of sheep for food. Wool is composed of more than 20 amino acids, which form long chains, or polymers, of protein. It also contains small amounts of fat, calcium and sodium. The coiled springs of wool’s molecular chains contribute to the fibre’s resilience. As it grows from the sheep’s skin, wool naturally groups into staples which each contain many thousands of fibres. There is considerable variation in the diameters of wool fibres, depending upon the type of sheep from which it comes, with the finer fibres being preferred for textiles.

2.2. Non-fibrous materials Commercial products manufactured from biomass materials in the form of various board products consisting of shredded and compressed wood, straw, reed or cork particles are already commercially available. Wassilieff [10] has already presented a comprehensive account of the acoustical characteristics of this type of material and hence it will not be examined in this paper. The main thrust of the investigation of the acoustical properties of non-fibrous materials will be concerned with unmodified straw and reeds. These materials have a long history of use for roofing and Anthony reports that a lifespan of 40–60 years can be achieved

Table 1 Data for natural fibres. Substance

Raw cotton Flax fibre Ramie Raw wool Jute carded fibre Wool batts Jute raw fibre Hemp batts Sisal fibre

Fibre diameter (lm) Mean

Standard deviation of fibre diameter (lm)

13.5 21.8 24.4 37.1 62.1 63 81.2 93.9 213

0.9 5.4 12.1 9.1 17.4 15.8 37.0 34.8 61.4

Density (kgm3) Matrix material

Bulk Sample

1530 1500 1500 1300 1370 1300 1370 1480 1410

40.5 78.4 96.1 19.8 49.1 25.7 65.6 75.9 38.6

Calculated porosity

Calculated flow resistivity (Pa s m2) Mechel-small radius

0.973 0.948 0.936 0.985 0.964 0.980 0.952 0.949 0.973

22,342 20,393 24,465

Mechel-large radius

Garai–Pompoli

26,306 30.057 1570 1956 842 1686 1361 112

43,370 26,705 29,863 1296 1743 680 1497 1398 106

352

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

by thatched roofs composed of reeds [9]. This history of durability is of importance. In the context of the Holiwood project and the requirement to devise a sustainable noise absorption treatment for a highway noise barrier. The pore distribution characteristics occurring in layers of unshredded straw or reeds are highly heterogeneous. In the case of straw and reeds, the pores arise from the non-uniformity of the cross sections of individual lengths. This results in a large number of slit like pores, parallel to the straw or reed lengths, which communicate with large cavities between the individual straws or reeds. In this paper we describe an investigation of the acoustical performance of a number of different orientations of reeds. 3. Parameters determining the performance of sound absorbent materials The intensity of plane wave reflected from an absorbent surface is smaller than that of the incident wave by a factor of |R|2 where R is the sound pressure reflection coefficient. The absorption coefficient, a, is thus given by:

a ¼ 1  jRj2

ð1Þ

It should be noted that the validity of the Delany–Bazely model is normally considered to be restricted to the range 0.01 < qof/ r < 1. These limits relate primarily to the samples employed for their experiments, however, Delany and Bazely state that there are theoretical reasons for expecting the data to normalise to other power-law relationships outside this range. The flow resistivity is determined by the porosity and pore size, and the latter is dependent on the size and shape of the fibres. The Delany–Bazely relationships were obtained from measurements on a range of glass and mineral fibre materials for which the fibre diameters were typically 1–10 lm. Garai and Pompoli [8] suggested that the Delany–Bazely model was unsuitable for predicting the acoustical characteristics of polymer fibres for which the diameters ranged from approximately 20 to 50 lm. They carried out a similar set of tests on polymer fibres to those of Delany and Bazely to obtain the following expressions for predicting their acoustical characteristics:



qf zc ¼ qo c 1 þ 0:078 o r

0:623

 0:660 ! qf  j0:074 o

r

 0:53  qf qf  j0:159 o kc ¼ x=c 1 þ 0:121 o

r

r

ð6Þ

0:571 ! ð7Þ

The reflection coefficient is given by:

R¼

zs  q0 c zs þ q0 c

ð2Þ

where zs is the surface impedance of the surface and p0c is the characteristic impedance of the propagating medium. In the case of a single layer of porous absorber of depth d with a rigid backing the surface impedance zs is given by:

zs ¼ jzc cotðkc dÞ

ð3Þ

where the zc is the characteristic impedance and kc is the complex wave number of the absorptive material. Thus it is the thickness of the sample and the complex impedance and complex wave number of the material that determine its absorptive characteristics. 3.1. Fibrous materials Conventional porous sound absorbers come in a variety of forms and their performance depends upon their pore structure. The most common form of porous sound absorbers are highly porous and homogeneous in structure and consist of fibrous materials and open cell foams. Fibrous materials constitute a large proportion of porous sound absorbers in common use and a number of empirical models have been developed which build upon the work of Delany and Bazley [7], presented in Eqs. (4) and (5), which employ the single nonacoustical parameter of flow resistivity for predicting their acoustical characteristics.

zc ¼ qo c 1 þ 0:0571



kc ¼ x=c 1 þ 0:0978

qo f r



0:754

qo f r

 0:732 ! qf  j0:087 o

r

0:7

 0:595 ! qf  j0:189 o

r

Garai and Pompoli do not provide any information regarding the range of validity of their expressions in the same form as that given by Delany and Bazely. Examination of the sample data that they provide indicate that the power-law relationships were devised for the range 0.05 < qof/r < 8.4. However, it can be seen on the figures that they provide that the agreement between the measured data and the predicted absorption coefficients obtained using Eqs. (6) and (7) are relatively poor at the very low and high frequencies corresponding to the extremes of this range. The Garai–Pompoli model is of particular interest in the context of this work as the diameters of the fibres and the density of the matrix material of polymer fibres are both closer to those of natural fibres than glass or mineral fibres. 3.1.1. Calculation of flow resistivity Both of these models require information concerning the flow resistivity of the material. Equations for predicting the flow resistivity of fibrous absorbers based upon fibre radius and material porosity or the bulk density of the materials have been presented by Mechel [11] Bies and Hansen [12] and Garai and Pompoli [8]. Mechel [11] gives the following relationships for sound incidence perpendicular to the direction of the fibres:

r¼

ð5Þ

where qo is the density of air, r is the flow resistivity, f is the frequency and x = 2pf is the angular frequency. Delany and Bazley obtained simple power-law relations by best-fitting a large amount of experimental data for a range of fibrous porous absorbers. Although a good empirical match was achieved, these relationships are only applicable over a well defined frequency range and where the porosity, e, which is the volume fraction occupied by pores in the material, is close to 1.

ð8Þ

6:8gð1  eÞ1:296 a2 e3

ð9Þ

and

r¼ ð4Þ

10:56gð1  eÞ1:531 a2 e3

where g is the viscosity of air (equal to 1.84  105 Pa s), a is the radius of the fibres and e is the porosity. Eq. (8) relates to fibre diameters ranging from 6 to 10 lm and Eq. (9) relates to fibre diameters from 20 to 30 lm. For a material having only a small amount of binder and assuming the presence of no closed cells, the porosity, e, is given by:

e  1  q=qm

ð10Þ

where q is the bulk density of the material and qm is the density of the matrix material. For materials with the same (or very similar) matrix materials, as can be seen from Eq. (10), the bulk density will be a function of

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

the porosity and Bies and Hansen [12] have presented the following expression which relates the flow resistivity to the bulk density, q, of the porous material.

r¼

3:18  109 4a2 q1:53

ð11Þ

The Bies and Hansen expression, Eq. (11), was obtained from measurements on fibre glass and mineral fibres and the density of the matrix material in both cases is approximately 2600 kg m3. Garai and Pompoli [8] made similar measurements of the flow resistivity of a number of polymer fibres for which the fibre diameter varied between 18 and 48 lm with a mean value of 33 lm and for which the density of the matrix material was approximately 1350 kg m3. They obtained the following expression for flow resistivity:

r¼

28:3  109 4a2 q1:404

ð12Þ

The Mechel, expressions (Eqs. (8) and (9)) were derived for fibres with diameters from 6 to 10 and 20 to 30 lm and the Bies and Hansen equation (Eq. (11)) was derived for mineral wool and fibre glass fibres with diameters from 1 to 10 lm. Examination of Table 1 reveals that the diameters of most natural fibres are considerably greater than those of mineral wool or fibre glass apart from cotton, flax and ramie which have mean diameters of 13.5, 21.8 and 24.4 lm, respectively. Thus Eq. (8) can potentially be used for cotton and Eq. (9) can potentially be used for flax and ramie. The Bies and Hansen expression, Eq. (11), in which the bulk density is used instead of porosity to predict flow resistance, was obtained from measurements on fibre glass and mineral fibres. The density of the matrix material in both cases is much greater than that of typical biomass fibres. The density of the matrix material of natural fibres is generally in the range 1300–1500 kg m3 which is closer to that of the polymers investigated by Garai and Pompoli, approximately 1350 kg m3, than to that of glass or minerals which are both typically approximately 2600 kg m3. Thus, it might be expected that the Gari and Pompoli expression for predicting the flow resistivity of fibrous samples might be applicable to natural fibres. Calculated values of flow resistivity for the natural fibres examined in this work are presented in Table 1. The Garai–Pompoli expression is used for all materials whilst the Mechel equation for fibres with small diameters is employed for fibres with diameters less than 14 lm and the Mechel equation for fibres with large diameters is employed for fibres with diameters greater than 22 lm. The range of flow resistivities calculated using the Garai– Pompoli model is from 106 to 59,830 Pa s m2. However, the range examined in their work only extended from approximately 900 to 23,000 Pa s m2 and hence their equation may not be valid outside this range. The range of flow resistivities considered by Delany and Bazel was from 1000 to 50,000 Pa s m2 and hence their equation may not be valid outside this range. The flow resistivity of the cotton samples predicted using the Garai–Pompoli equation is almost twice that obtained using the Mechel equation. This difference is due in part to the higher density of the matrix material of the cotton which differs significantly from that of the polymers investigated by Garai and Pompoli than the other fibres. In addition, the diameters of all of these fibres are outside the range on which the Garai–Pompoli equation was based. Thus it can be concluded that use of the Mechel small diameter equation is more appropriate for predicting the flow resistivity of these fibres. From examination of Table 1 it can be seen that the mean diameters of the flax and ramie fibres fall within the range of validity of both the Mechel large diameter equation (20–30 lm) and the Garai and Pompoli equation (18–48 lm). It can also be seen that the flow

353

resistivities calculated by both equations are in very close agreement for these fibres and hence both are equally appropriate. With regard to the remaining materials, only mean diameter of the raw wool falls within the range of validity of the Garai and Pompoli expression. However, both the carded jute and wool batts have mean diameters that are relatively close to the upper limit of validity of the Garai–Pompoli expression, that is 62.1 and 63 lm, respectively, compared to 48 lm. The diameters of the remaining fibres are well above this upper limit. The Mechel large diameter expression and the Garai–Pompoli expression predict similar values for the coarser fibres but the former tends to yield values that are 5–10% higher. 3.1.2. Application of predictive models to natural fibres In this work the effectiveness of the empirical models for predicting the acoustical characteristics due to both Delany and Bazely, Eqs. (4) and (5), and Garai and Pompoli, Eqs. (6) and (7), is investigated when applied to biomass fibres. In both cases the model due to Garai and Pompoli, Eq. (12), was employed to calculate the flow resistivity for all fibre samples in view of the similarity between the densities of the matrix materials of the biomass and polymer fibres. In addition, where the fibre diameters were within or close to the ranges of applicability of the Mechel expressions as discussed in Section 3.1.2, Eqs. (8) and (9), these were also employed. 3.2. Non-fibrous materials The non-fibrous absorbers investigated in this paper consisted of straw and reed configurations. Their pore structure differs from most porous absorbers as it arises from the space between individual stems and the non-uniformity of the cross sections of individual stems. This results in a large number of slit like pores, parallel to the straw or reed lengths, which communicate with large cavities between the individual straws or reeds. Models for predicting the characteristic impedance and wave number of porous materials with structures which are more complex than fibres are generally developed from considerations of viscous and thermal effects on the sound propagation in circular pores and between parallel plates. Models for these materials thus require more detail to account for the interaction between the sound waves and the pores of the material. The micro-structural approach for arriving at models for predicting the acoustical characteristics of such materials, as employed Zwikker and Kosten [13], involves deriving the wave propagation inside individual pores from first principles and then generalising the results to the macroscopic scale. The viscous and thermal effects are separately leading to the concepts of complex density, qb, and complex compressibility, Cb, from which the characteristic impedance and complex wave number can be calculated as follows:



1

qb ðxÞ 2 C b ðxÞ 1 kc ¼ jx½qb ðxÞC b ðxÞ2

zc ¼

ð13Þ ð14Þ

Predictive models for materials with complex pore structures require a number of additional parameters such as tortuosity, pore shape factor and the standard deviation of the pore size distribution. The tortuosity, Ts, takes into account the orientation of the pores relative to the incident sound wave which also has an effect on the sound propagation in the pores. This parameter cannot be calculated accurately and in practice Ts is measured. The pore shape factor, sf, takes into account the dependence of the propagation of sound in the pore on the shape of the pore, as

354

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

it affects the viscous and thermal interaction effects. For simple pore shapes it can be determined analytically. However for most absorbers the pore shapes are complicated and it has to be measured. Biot [14,15] and Smith et al. [16] introduced a dynamic shape factor to allow for a pore shape with a range of cross sections. For application to bulk media, Dupuit relationships [17] were applied using a modification which included the effect of tortuosity. Other models have been developed by Attenborough [18], Stinson [19], and Allard and Champoux [20] which all use flow resistivity, porosity, tortuosity and the shape factor to predict the performance of a porous absorber. Models for the characterisation of materials with a highly heterogeneous pore structure require the incorporation of a parameter to account for the pore size distribution in the material structure. Although models have been presented for predicting the acoustical characteristics of heterogeneous material structures [21], this has related to pore size variation at a microscopic scale. However, the pores associated with configurations of whole reeds or straws are relatively large and the applicability of any current theoretical model to this configuration is doubtful. For this reason the investigation of the acoustic performance of straw and reed configurations described below is empirical.

4. The measurement procedures Measurements of the sound absorption characteristics of a range of biomass materials were carried out using a combination of the impedance tube and reverberation chamber methods. The impedance tube measurements are based on the two-microphone transfer-function method according to ISO 10534–2 [22]. With this method it is possible to obtain fast measurements of normal incidence parameters using small samples that are easy to assemble/ disassemble. However, the results of a recent round-robin experiment have demonstrated that this method can give different results when employed in different laboratories [23], possibly due to differences in the preparation of samples. In general the deviation between results is greater for materials with a very high value of flow resistivity coupled with a large standard deviation in this parameter. The impedance tube was employed in this work for measurements on fibrous materials. The measurements presented in this paper were made using the Brüel & Kjær Impedance Tube Kit Type 4206. Sample holders of 100 mm and 29 mm in diameter are provided; the larger is employed for measurements over the range of frequencies from 50 Hz to 1.6 kHz and the latter over the range of frequencies from 500 Hz to 6.4 kHz. The impedance tube method has a number of advantages over the reverberation chamber method. First, the apparatus itself is much smaller and therefore more practical. Secondly, only a small sized sample is required for the tests, and thirdly it allows the surface impedance to be determined in addition to the absorption coefficient. The disadvantages are first, that the properties are only measured for sound at normal incidence to the sample although it is possible to apply a correction to obtain an approximate value of the random incidence absorption coefficient. Secondly, uncertainties are introduced when measuring heterogeneous materials as the constitution and pore structure of samples taken from different regions of a large sample may vary considerably. Thirdly, two different samples are required for measurements over a large frequency range. The impedance tube was employed in this work for measurements on homogeneous materials. Finally, when cutting samples out of sheet material errors can be introduced due to any circumferential air gaps between the sample and the tube

[24]. These errors are generally greater for materials with a high flow resistivity. The reverberation chamber method is the basis of EN 20354 [25] and measures the absorption properties of a material sound waves at random incidence. The method is based on measuring the reverberation time in a room before and after the introduction of the test samples. The absorption of the sample material is found by comparing the reverberation time measurements, taken without the sample in the room, to the reverberation time measurements taken with the sample in the room. In the reverberation chamber method sound arrives at the material from arbitrary angles and hence the measured absorption properties are more representative of the performance of the material under real conditions. However, the method requires large samples of the material under investigation as it is not possible to get accurate data from small samples. The edges of the sample constitute a possible source of error in this method as sound is diffracted at the edges of the sample which leads to excess absorption predictions. To reduce this problem, the usual practice is to cover the material edges with some form of frame. The reverberation chamber was employed to measure the absorption properties of different configurations of reeds. The reverberation chamber at the Acoustics Research Unit at the University of Liverpool measures 5 m  5 m  4.8 m and thus has a volume of 120 m3. EN 20354 [25] specifies a minimum volume of 150 m3 and therefore the Liverpool facility does not satisfy this requirement. However, in practice it is sufficiently close to the recommended value that measurements will be accurate apart from at the very lowest frequencies. 5. Results 5.1. Natural fibres Table 1 contains a summary of the sample materials examined. It can be seen that the plant based materials have very similar densities as they consist of the same basic principal components, cellulose, hemicellulose and lignin. Variations in the densities are due to different proportions of these components as their densities vary with values of 1397 and 1559 and 1520 kg m3 being reported for lignin, hemicellulose and cellulose, respectively [26]. Thus, the lower values of density for hemp and sisal result from a greater proportion of lignin than in the other fibres. Micrographs of the fibres were made and the mean fibre diameter was measured from the micrographs. Samples of the lighter coloured materials were dyed prior to preparing micrographs to aid visibility. The samples of loose fibres were hand compacted in the tube sample holders to give a sample thickness of 50 mm. The weight of the test sample was measured and an estimate of the porosity of the sample was obtained from Eq. (10). The density of the matrix material was obtained from published literature. After the bulk density of the samples for the large tube was measured the small tube samples were made to the same bulk density. The results of these measurements are summarised in Table 1. All measurements were made using the impedance tube method with both the large and small sample holders so that the absorption coefficient could be measured over the frequency range from 50 Hz to 6.3 kHz. 5.1.1. Basic fibres From examination of Eqs. (8)–(12), it is apparent that the factors that determine the absorption characteristics of fibrous absorbers are the fibre diameters and the porosity. The latter is a function of the degree to which the material is consolidated or compacted. In this work a variety of natural fibres were obtained and a systematic

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

355

investigation of their absorptive properties was carried out. The first stage involved an investigation of the acoustical characteristics of cotton, wool, jute and sisal. These materials were selected as their diameters differed considerably. Fig. 1 shows the measured absorption coefficient as a function of frequency for the raw cotton fibres. Cotton fibres are obtained from the boll or seed pod of the cotton plant. Once the seed pod has opened the raw cotton fibres dry out and can be harvested and made into textile fabrics and other cotton containing products. The raw cotton tested was unprocessed raw American cotton and it can be seen from Table 1 that it has the smallest average diameter, 13.5 lm, and the smallest standard deviation, 0.9 lm, of all the unprocessed fibres examined. It can be seen from Fig. 1 that the cotton fibre sample has absorption properties which are similar to those of rock wool or fibre glass of the same thickness. Also shown in Fig. 1 are the predicted absorption characteristics obtained using both the Delany and Bazely [7] and Garai and Pompoli [8] models. Both the Mechel [11] expression (Eqs. (8) and (9)) and the Garai and Pompoli expressions [8] (Eq. (12)) were employed to calculate the flow resistivity. It can be seen in Fig. 1a that the predictions using the Mechel expression for flow resistivity of small diameter fibres and the Delany–Bazely model for the acoustical characteristics are in very good agreement with the measured data. As shown in Fig. 1b, relatively poor agreement is observed with the combination of the Garai–Pompoli models for flow resistivity and acoustical characteristics. This may be in part due to the different values of flow resistivity calculated by the two methods, the Mechel expression yielding a value of 22,342 Pa s m2 and the Garai and Pompoli expression yielding a value of 43,370 Pa s m2. However, the predictions obtained using the Delany–Bazely model for the acoustical characteristics and the Garai–Pompoli model for flow resistivity are in slightly better agreement with the experimental data which suggests that the Garai–Pompoli model for the predictions of the acoustical characteristics is not valid for the small fibre diameters of cotton. With regard to the good agreement obtained with the combination of Mechel model for flow resistivity and Delany–Bazely model for the acoustical characteristics, it should be noted that both empirical models were

derived from measurements on fibre samples with similar diameters to the cotton. Fig. 2 shows the measured absorption coefficient as a function of frequency for the wool fibres. Raw wool was used for these tests, having a mean diameter of 37.1 lm. The measured diameters of the sample fibres were larger than those normally used for clothing where diameters less than 25 lm are preferred. The bulk density of the sample was low which reflects the difficulty experienced in hand compacting due to the springy nature of this fibre. It can be seen that the wool fibres act as very effective sound absorbers but are not as good as the cotton. This may be due in part to the relatively high porosity of the wool due to the difficulty in compacting it. Also shown in Fig. 2 are the predicted absorption characteristics obtained using both the Delany and Bazely [7] and Garai and Pompoli [8] models. Both the Mechel expression (Eq. (9)) and the Garai and Pompoli expressions (Eq. (12)) were employed to calculate the flow resistivity. The Mechel expression yielded a value of 1570 Pa s m2 and the Garai and Pompoli expression yielded a value of 1296 Pa s m2. As can be seen in Fig. 2b, the agreement between the measured and predicted data was found to be particularly good when the Garai and Pompoli expressions are used to calculate both the flow resistivity and the absorption characteristics. It should be noted that mean fibre diameter and density of the matrix material of the wool, 37 lm and 1300 kg m3, respectively, are very close to those of the polymers, 33 lm and 1350 kg m3, investigated by Garai and Pompoli. Fig. 3 shows the measured absorption coefficient as a function of frequency for the carded jute fibres. Jute is a coarse bast fibre with very long fibre lengths. The term carding refers to the use of a comb device to reduce the variability in fibre diameters and align the fibres in preparation for spinning. It can be seen that the absorption characteristics of the jute sample are very similar to those of the wool sample even though the fibres have a larger diameter than those of the wool. This is due to the difference in porosity between the two samples as the jute was easier to compact and thus the measurement sample had a lower porosity. Also shown in Fig. 3 are the predicted absorption characteristics obtained using both the Delany and Bazely [7]

Fig. 1. Comparison of predicted and measured values of normal incidence absorption coefficients of cotton fibre sample. (a) Prediction obtained with Delany–Bazely model: + experimental measurements; ........ flow resistivity predicted using Eq. (8); ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12). (b) Prediction obtained with Garai–Pompoli model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12).

Fig. 2. Comparison of predicted and measured values of normal incidence absorption coefficients of wool fibre sample. (a) Prediction obtained with Delany– Bazely model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12). (b) Prediction obtained with Garai–Pompoli model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12).

356

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

Fig. 3. Comparison of predicted and measured values of normal incidence absorption coefficients of jute fibre sample. (a) Prediction obtained with Delany– Bazely model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12). (b) Prediction obtained with Garai–Pompoli model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12).

Fig. 4. Comparison of predicted and measured values of normal incidence absorption coefficients of sisal fibre sample. (a) Prediction obtained with Delany– Bazely model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (11). (b) Prediction obtained with Garai–Pompoli model: + experimental measurements; ___ flow resistivity predicted using Eq. (9); - - -; flow resistivity predicted using Eq. (12).

and Garai and Pompoli [8] models. Both the Mechel expression (Eq. (9)) and the Garai and Pompoli expressions (Eq. (12)) were employed to calculate the flow resistivity. The Mechel expression yielded a value of 1956 Pa s m2 and the Garai and Pompoli expression yielded a value of 1743 Pa s m2. These values are similar in magnitude to those for the wool sample and hence the predicted sound absorption characteristics are similar. Although the measured data for the jute sample tend to follow the general trends of the predicted absorption characteristics obtained by application of both models, the agreement is not good. The measured values are greater than the predicted values for both models. The predictions obtained using the Delany–Bazely model and Mechel equation for predicting the flow resistivity are in slightly better agreement with the measured data at frequencies above 1 kHz, as can be seen in Fig. 3a. Fig. 4 shows the measured absorption coefficient as a function of frequency for the sisal fibres. Sisal fibre is derived from the leaves of the sisal plant. It is usually obtained by machine decortications in which the leaf is crushed between rollers and then mechanically scraped. The fibre is then washed and dried by mechanical or natural means. The strands average from 80 to 120 cm in length and 200 lm to 400 lm in diameter. The mean diameter of the sample fibres employed for these tests was 213 lm. It can be seen from Fig. 4 that this is a relatively poor sound absorber. This is due to the very coarse nature of the fibres and difficulty in compacting them in the sample holder resulting in a high value of porosity. Also shown in Fig. 4 are the predicted absorption characteristics obtained using both the Delany and Bazely [7] and Garai and Pompoli [8] models. Both the Mechel expression (Eq. (9)) and the Garai and Pompoli expressions (Eq. (12)) were employed to calculate the flow resistivity. The Mechel expression yielded a value of 112 Pa s m2 and the Garai and Pompoli expression yielded a value of 106 Pa s m2. These values are well below the lowest value of 880 Pa s m2 recorded by Garai and Pompoli in their experiments and hence possibly beyond the range of validity of Eq. (12). Similarly, the mean fibre diameter of 213 lm is well above that for which the Mechel expression of Eq. (9) is applicable. Thus the predicted values of flow resistivity are likely to be unrealistic. Similar reservations regarding the applicability of both the Delany–Bazely and Garai–Pompoli models for predicting the

absorption coefficients of the samples apply. Thus, although the trend of the measured data for sisal shown in Fig. 4 follow the form of the predictions of both models, it is not surprising that the agreement is very poor with the measured values being consistently greater than the predicted values. As can be seen from Fig. 4a and b, the predicted values are only slightly affected by the choice of model for the prediction of flow resistivity or the choice of model for predicting the acoustical characteristics. The combination of the Mechel large diameter expression for predicting the flow resistivity and the Delany–Bazely model for predicting the absorption coefficient yields marginally better agreement with the measured data. It can be concluded that existing prediction models can produce results in close agreement with measured data for biomass fibres with diameters less than approximately 60 lm. However, they have limited applicability when dealing with most natural fibres where fibre diameters are relatively large. This is probably due to the diameters of these fibres differing considerably from those on which the predictive models have been developed and the possibility that the surfaces of the very coarse fibres might also contain micro-pores. Comparison of the values of flow resistivity calculated using the large diameter Mechel expression based upon porosity and the Garai–Pompoli expression based upon bulk density showed that the two methods resulted in similar values apart from the case of fibres with diameters of the order of 10 lm where the Garai–Pompoli predicts larger values of flow resistivity than the small diameter Mechel expression. The latter expression, when employed with the Delany–Bazely model for predicting the acoustical characteristics of fibres with small diameters yields the most accurate predictions. For the larger diameter fibres examined, the large diameter Mechel expression predicted slightly higher values of flow resistivity than the Garai–Pompoli expression and the use of this value with the Delany–Bazely model generally resulted in the most accurate predictions of the acoustical characteristics of a sample. However, this was not the case with the wool sample tested where the combination of the Garai–Pompoli expressions for flow resistivity and acoustical characteristics yielded the best agreement. It should be noted that there is a degree of uncertainty associated with the predicted values due to the variability of the diameters of

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

individual fibres in each sample as shown by the values of the standard deviation given in Table 1. From the above results it can be concluded that in addition to cotton the most suitable natural fibres for use as porous sound absorbers are those for which the average diameters are small and which are capable of being well compacted. From examination of Table 1 the most promising materials that can be identified are flax and ramie which have average diameters of 21.8 lm and 24.4 lm, respectively. Flax is a bast fibre with a long history of use for clothing due to its relatively small diameter when compared with hemp and jute. Ramie is one of the oldest fibre crops, having been used for at least six thousand years, and is principally used for fabric production. It is a bast fibre, and the part used is the bark of the vegetative stalks. Fig. 5 shows a comparison between measured and predicted data for flax and ramie fibres. Predictions were made using both the Delany and Bazely and Garai–Pompoli models with the value of flow resistivity calculated using the Mechel expression (Eq. (9)). It can be seen from Table 1 that the Mechel large diameter equation and the Garai–Pompoli equation results in very similar values of flow resistivity for these fibres and hence the choice of equation is not important. Both the flax and ramie fibres exhibit very good sound absorption characteristics at high frequencies. The predictions of the Delany–Bazely model are in better agreement with the measured data than those of the Garai–Pompoli model which tends to predict higher values of absorption coefficient than those measured. The agreement is particularly good for the ramie data. 5.1.2. Processed fibres Natural fibres can undergo a number of processes such as carding in which the raw fibres are aligned by a combing process and fibres with large diameters removed. Samples of both raw jute and carded jute fibres were available for testing. The raw material had a slightly greater mean fibre diameter, 82.2 lm compared to 62.1 lm, and higher bulk density which reflects the greater difficulty in compacting the more ordered bundles of fibres arising from the carding process. Fig. 6 shows the measured data for both raw and carded jute fibres. It can be seen that effect of carding is to reduce slightly the absorption coefficient at most frequencies. Also shown in Fig. 6

Fig. 5. Comparison of predicted and measured values of normal incidence absorption coefficients for flax and ramie fibre samples. (a) Flax: + experimental measurements; ____ predicted values using Delany–Bazely model; - - - - predicted values using Garai–Pompoli model. (b) Ramie: + experimental measurements; ____ predicted values using Delany–Bazely model; - - - - predicted values using Garai– Pompoli model.

357

Fig. 6. Comparison of predicted and measured values of normal incidence absorption coefficients for raw jute and carded jute fibres. (a) Raw jute:: + experimental measurements; ____ predicted values using Delany–Bazely model; - - - - predicted values using Garai–Pompoli model. (b) Carded jute: + experimental measurements; ____ predicted values using Delany–Bazely model; - - - - predicted values using Garai–Pompoli model.

are the absorption coefficients predicted using the Garai–Pompoli expression for flow resistance and both the Delany–Bazely and Garai–Pompoli models for calculating the characteristic impedance. It can be seen that although both models predict the form of the absorption characteristics of the jute fibres they tend to under predict the values of the absorption coefficient, particularly at the higher frequencies. The differences are greater for the coarser jute fibres. In both cases the Delany–Bazely predictions are in better agreement with the measured data than those of the Garai– Pompoli model. The differences between the measured and predicted data may be due to the diameters of the fibres being outside the range involved in the derivation of both the Delany–Bazely and Garai– Pompoli models. It can be seen from Fig. 4 and the data relating to the very coarse sisal fibre that the difference between measurements and predictions is very large. This lends support to the hypothesis that the existing models cannot be applied to fibres having very large diameters. A number of natural fibres are sold in the form of batts in which some form of binding agent is employed to hold the material in the form of a thick rectangular slab which can be easily handled. These batts are used for thermal insulation and also as an acoustic absorber for use in stud walls. Biomass materials used in the form of batts include wool, flax and hemp. Hemp is a coarse bast fibre with very long fibre lengths hence its use in ropes and twine. Hemp was available in the form of hemp mats and hemp batts. The former consisted of thin sheets of highly compacted fibres and was intended for use as animal bedding. The hemp batts were less highly compacted and consisted of 85% hemp fibres bound into a ‘batt’ with approximately 15% of recycled polyester to act as a binder. The polyester is mixed in with the hemp then heated to soften it and fuse the batt together on cooling. The polyester was ignored for the purposes of estimating the fibre size although they could be seen on the micrograph as rather large inclusions. Wool was available in two forms: firstly as unprocessed sheep wool, breed of sheep unknown, and also in the form of a wool batt primarily marketed as a thermal insulation batt but also as an acoustic absorber for use in stud walls. The wool batt also incorporates an unspecified amount of recycled polyester, employed as a binder.

358

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

It can be seen from Fig. 8a and b that for both configurations the samples do not fit tightly against the walls of the sample holder. However, although Pilon et al. [24] have pointed out the problems that can arise from circumferential gaps in impedance tube sample holders when working with high flow resistivity material samples cut from sheets, it is probable that the effect experienced with these coarse, large pored samples will differ. This is discussed further in the following section.

Fig. 7. Comparison of predicted and measured values of normal incidence absorption coefficients for wool and hemp batts. (a) Wool batt: + experimental measurements; ____ predicted values using Delany–Bazely model; - - - - predicted values using Garai–Pompoli model. (b) Hemp batt: + experimental measurements; ____ predicted values using Delany–Bazely model; - - - - predicted values using Garai–Pompoli model.

Fig. 7 shows the measured absorption characteristics for the hemp and wool batts. Also shown in Fig. 7 are the absorption coefficients predicted using the Garai–Pompoli expression for flow resistivity and both the Delany–Bazely and Garai–Pompoli models for calculating the acoustical characteristics. It can be seen that whilst both models predict the form of the absorption characteristics they tend to under predict the actual values of the absorption coefficients at the higher frequencies. There are two possible reasons for this. The first is that the mean diameters of the fibres, 63 lm and 93.9 lm for the wool and hemp batts, respectively, were considerably beyond the upper limits of ranges examined by both Delany and Bazely and Garai and Pompoli. The difference is particularly great for the hemp fibres which again lends support to the hypothesis that the models are not applicable to fibres of large diameters. In addition the presence of a significant amount of binder will affect the bulk density and possibly the homogeneous nature of the samples. 5.2. Non-fibrous materials There are already examples of biomass materials employed in the manufacture of sound absorbent materials. They may be used in shredded form, either bound together by means of a process which releases natural sugars that then act as a glue or, for a more durable solution, bound together with small amount of cement. Alternatively the biomass material can be reduced to its basic fibres which are then compressed and bound together in the form of a board product. Wassilieff [10] has presented the results of a detailed study of such materials and demonstrated how their properties may be predicted using Attenborough’s model [18] hence they will not be considered in this paper. However, an alternative will be examined consisting of aligned lengths of whole reed or straw stems. Two basic configurations were examined, the first was end-on configuration in which the cut ends of the material are perpendicular to the incident sound, as shown in Fig. 8a. The second was the transverse configuration in which the reed stems are perpendicular to the incident sound as shown in Fig. 8b. Because of the irregular nature of the reed cylinder there are slit-like gaps between reeds as can be seen in Fig. 8d and these will link through to the large voids between reeds that are visible in Fig. 8c.

5.2.1. Straw and reeds in the end-on configuration The ‘‘end-on’’ structure consists of straw or reeds separated by large prismatic voids. The straw and reeds are hollow tubes with a small internal diameter and it is to be expected that there will be dissipative losses as sound propagates in these tubes. The tubes also contain pith which might be expected to contribute to the sound absorption. However, the holes in the straw and reed are blind, being closed at various points along their length. Although the cylindrical holes might be expected to make a significant contribution to the sound absorption, it is not clear how they might affect the bulk flow resistivity. Reed and straw arranged in this orientation do not conform well to heterogeneous models such as that developed by Horoshenkov et al. [21]. Thus no attempt has been made to try to investigate the applicability of any existing models to the prediction of the acoustical performance of straw or reed assemblies. Fig. 9 shows the measured absorption coefficient of bundles of straw and reed 50 mm long as a function of frequency for the ‘‘end-on’’ configuration. For these measurements both the small and large diameter impedance tubes were employed to permit measurements to be made over a wide frequency range. The absorption characteristics of the reeds and the straw are similar but the straw is more effective at frequencies above 1 kHz, possibly because of greater frictional losses. For both materials well defined peak values of absorption coefficient can be seen at a frequency of approximately 1250–1600 Hz and 5000 Hz with an equally well defined trough at a frequency of approximately 3150 Hz. A length of 50 mm corresponds to a quarter wavelength for a frequency of approximately 1600 Hz, three quarter wavelengths at approximately 4800 Hz and a half wavelength at approximately 3300 Hz. Thus it may be surmised that the observed characteristics are due to resonance effects. This was investigated by repeating the measurements with reeds of lengths 8.5 cm, 10 cm and 15 cm. The measured data is shown in Fig. 10. It can be seen from Fig. 10 that the effect of increasing the length of the reed stems was to reduce the frequencies at which the first peak and the first trough in the absorption characteristics occurred. Simple calculations revealed that the relationships between the length of the stems and the wavelengths corresponding to the frequencies of the peaks and the troughs, as discussed above, held for the new reed lengths. Measurements of these reed configurations were only made with the larger tube of the impedance tube measuring system as the size of the reed stems (typically around 5 mm in diameter) was such that there was potentially considerable variation in the porous nature of different samples when used with the 29 mm diameter tube required for measurements at the higher frequencies. As the frequency at which the absorption peak corresponded to the reed length equalled approximately one quarter wavelength it would appear that either the hollow reeds or the cavities between reeds were acting as a quarter wave resonators. The hollow ends of the 8.5 cm reeds were sealed with putty so that only the cavity remained active and the measurements repeated. These results are also shown in Fig. 10. It can be seen that the effect of sealing the reeds is to increase the absorption coefficient slightly at most frequencies and also to lower slightly the frequency at which the peak

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

359

Fig. 8. (a) End-on reeds in large sample holder. (b) Transverse reeds in large sample holder. (c) Magnified view of end-on reeds showing pith in reed tubes and gaps between reeds. (d) Magnified view of transverse reeds showing gaps between aligned reeds.

Fig. 9. Normal incidence absorption coefficient of straw and reeds in end-on configuration measured using small and large impedance tubes.

Fig. 10. Normal incidence absorption coefficient of end-on reeds of different lengths measured using large impedance tube.

absorption occurred. It can be concluded that the narrow space between the reed stems is the most important feature of this configuration with regard to its sound absorption characteristics. As can be seen in Fig. 8a, the size of voids between the reeds and the walls of the sample holder are comparable to the voids between neighbouring reeds, the circumferential effect discussed by Pilon [24] et al. is unlikely to be significant . As stated above, given the relative diameters of the sample holder and the reed stems it is probable that there are potentially considerable variation in the porous nature of different samples even when using the large sample holder. Nevertheless, the results do demonstrate the potential of end-on configurations of reeds and straws as sound absorbers and suggested that measurements on larger samples in a reverberation chamber would be valuable.

Fig. 11 shows the results of measurements made with 12 m2 sample of end-on reeds cut to an approximate length of 14 cm and bordered by a 14 cm high wooden frame. It can be seen that this reed configuration was very effective at absorbing sound at frequencies above 250 Hz. The expected peak at around 600 Hz, the frequency for which 14 cm corresponds to the quarter wavelength, cannot be observed. This is probably due to the sound incidence being random in nature rather than perpendicular to the sample and also because of the variations in the lengths of the reed stems leading to a broadening of the frequency response. The variability in the length of the reed stems arose from the difficulty experienced in cutting the large number of reeds required (approximately 500,000) to make such a large sample and

360

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

5.2.2. The transverse configuration The measured values of the absorption coefficient as a function of frequency for the transverse configuration which consisted of layers of reeds and straw perpendicular to the incident sound field as measured in the impedance tube are shown in Fig. 12. Measurements of these reed configurations were only made with the larger tube of the impedance tube measuring system as the size of the reed stems was such that there was potentially considerable variation in the porous nature of different samples when used with the 29 mm diameter tube required for measurements at the higher frequencies. It should be noted that obtaining consistent results from impedance tube measurements with transverse samples of reed and straw proved particularly difficult as there was a need to seal the

edge of the sample to avoid the circumferential effect discussed by Pilon et al. [24]. It can be seen that the values of absorption coefficient are small at low frequencies, rising with increasing frequency but exhibiting significant peaks. The results show a pronounced peak in the absorption coefficient at 630 Hz for reeds and 800 Hz for straw. The different value of the frequency at which the maximum absorption coefficients are observed is probably due to the difference in the diameters of reed and straw stems. The impedance tube measurements have indicated that reeds in the transverse configuration have some potential for application as a sound absorbent treatment. However, the small samples employed in the impedance tube cannot be assumed to accurately replicate the pore structure to be found in large areas and hence reverberation room measurements were essential. The reeds were delivered in the form of 1 m  1 m mats with a thickness of 50 mm and in the form of 1 m wide rolls of thickness 35 mm. Samples of area 12 m2 were made up using combinations of mats and rolls with thicknesses of 35 mm, 50 mm, 100 mm and 150 mm. The random incidence sound absorption coefficients measured in the reverberation chamber are shown in Fig. 13. The double and triple layers of mats were both arranged with all mats aligned in the same direction as not only was this configuration found to give a slightly better performance but it is probably the most practicable for manufacturing purposes. Comparison of the data for the 5 cm thick reed layer shown in Fig. 13 with the data for the 5 cm thick reed sample employed for the impedance tube measurements shown in Fig. 12 shows some similarities. In particular, both sets of data exhibit a pronounced peak in the mid frequency range, at 630 Hz for the impedance tube sample and 500 Hz for the reverberation room sample. This suggests that it may be possible to obtain meaningful data from measurements on small samples of different reed configurations by means of the impedance tube technique. Differences between the two sets of data may be due to the nature of the different incident sound fields, random incidence and normal incidence, respectively. At low frequencies it can be seen that the effect of increasing the thickness of the reed layer is to move downwards the frequency at which the absorption coefficient peaks and to introduce additional peaks. There is a slight increase in the peak low frequency absorption coefficient for the 150 mm thick configuration. At high frequencies there is a slight increase in the absorption coefficient with increasing layer thickness. The increase in

Fig. 12. Absorption coefficient of transverse straw and reeds (aligned perpendicular to the incident sound) measured using large impedance tube.

Fig. 13. Absorption coefficient of aligned reeds of different thickness measured in reverberation room.

Fig. 11. Absorption coefficient of end-on reeds of nominal length 14 cm measured in reverberation room.

this was eventually achieved by cutting up 5 cm thick, wire bound reed mats into strips. The need to maintain the wire binding in place so that the samples could be handled restricted the dimensions of the strips into which the reeds could be cut. In addition, the use of a band saw to cut these very flexible mats adversely affected the consistency of the cutting procedure.

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

361

absorption coefficient is greater for an increase from 50 mm thick to 100 mm thick than for an increase from 100 mm thick to 150 mm thick. This trend suggests that it is probable that there would be an additional increase if the thickness were to be made even greater but this would small. 5.2.3. Composite absorbers The results of the previous sections have shown that many natural fibres can be used to provide high frequency absorption and that highly heterogeneous porous materials in the form of aligned reeds can provide very good low frequency and moderate medium and high frequency sound absorption. This reflects the situation with conventional sound absorbers where fibrous materials such as mineral wool provide high frequency absorption and other systems, notably panels over air spaces, provide low frequency absorption. In some circumstances, such as in broadcast studios, a considerable amount of sound absorption is required over the entire audio frequency range and the limited area available for the application of treatment results in the use of composite absorbers [27]. These typically consist of a limp panel over an air space covered with a thick layer of porous absorber and the depth of such treatment can be considerable. The performance of the aligned reeds at low frequencies is similar to that of panel absorbers hence the effect of employing a thick reed underlay topped with a porous absorber was investigated. Fig. 14 shows the results measured in the reverberation chamber of the absorption coefficients for absorber systems consisting of 5 and 10 cm layers of reeds under a 7 cm thick hemp batt. It can be seen that the good low frequency performance of the reeds, especially for a thickness of 10 cm, is maintained but complemented by the good high frequency performance of the hemp. The result is a composite absorber which is relatively shallow but comparable in performance with the very specialised systems employed in broadcast studios [27]. One of the problems with using biomass fibres for acoustic absorbers is the need to bind the fibres together without the use of unsustainable substances. An alternative to adhesive binding is to use a mechanical system to hold the fibres in place. As there are gaps between aligned reeds due to irregularities in their sections, it is possible that a reed layer placed in front of fibrous material could act as a perforated protective panel and ensure dimensional stability. Measurements were therefore made of a composite system composed of 10 cm thick reed underlay, 7 cm

Fig. 14. Absorption coefficient of broadband composite sound absorber consisting of reed underlay with hemp batt on top measured in reverberation room.

Fig. 15. Absorption coefficient of composite system composed of 10 cm thick reed underlay, 7 cm hemp batt and thin double reed surface measured in reverberation room.

hemp batt and thin double reed surface. The latter consisted of two layers of a very thin, single reed roll. The results are shown in Fig. 15 along with the absorption coefficient of the 10 cm thick reed underlay topped with the 7 cm hemp batt. It can be seen that the reed screen, being acoustically transparent, allows the sound to reach the hemp and as a result the absorption coefficient of the composite arrangement is very similar to that of the simple 10 cm thick reed underlay and 7 cm hemp batt.

6. Conclusions The suitability of biomass materials as novel and sustainable sound absorbent treatments has been investigated. The biomass offers many options including fibres, wood based products and different configurations of straw and reed. It has been found that existing models for predicting the acoustical characteristics of fibrous absorbers produce data in reasonably good agreement with that obtained from measurements for fibres with diameters less than approximately 60 lm. The model due to Delany and Bazely was found to predict values of absorption coefficients for fibres with a large range of diameters that were in better agreement with measured values than predicted by the model of Garai and Pompoli. However, the latter model gave more accurate predictions for the case of wool for which the fibre diameters were similar to those of the polymer fibres on which the Garai–Pompoli model was based. Both models were not effective when dealing with large diameter fibres. This failure may be due to the limitations of the current models for predicting the acoustical characteristics and/or the flow resistivity due to the differences in the diameters of the fibres involved in their derivation from those of the coarser natural fibres. This could be resolved by a systematic study similar to that carried out by Delany and Bazely [7] and Garai and Pompoli [8]. The use of natural fibres offers a way of producing very absorbent materials whose acoustical absorption properties are determined by the typical fibre diameter and the degree of consolidation. Impedance tube measurements of the sound absorption characteristics of natural fibres having small diameters has confirmed their effectiveness as porous sound absorbers with properties similar to those of conventional absorbers made from rock wool or fibreglass. The general conclusion can be drawn that the most promising natural fibres for use as porous sound absorbers

362

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363

are those for which the average diameters are small and which are capable of being well compacted. In practical applications, however, other performance criteria have to be achieved. Of particular importance in situations in which porous absorbers might be employed are durability and fire resistance. The biomass materials examined in this work have been largely composed of cellulose and are not generally attractive as food to insects. However, if exposed to water they can be subject to fungal attack. In a dry building this will present no problem but might preclude their application in some environments. With regard to fire risk, cellulose based materials such as hemp thermal insulation and recycled newspapers used as thermal insulation, are routinely treated with fire retardant chemicals to make them safe and hence this problem can be overcome. Another practical consideration is the need to develop suitable binders to hold the fibres together without adversely affecting the absorption characteristics. A binder is needed both to reduce the possibility of a material shedding fibres and also to ensure that it maintains its shape. For example, in this work it was found that the natural springiness of wool tended to limit how much it could be compressed and this affected the resulting porosity. A suitable binder could hold the material in a more compressed form. In addition, fibrous materials without a binder would tend to move over time due to the effect of gravity. Straw and reeds have been found capable of acting as efficient sound absorbers. Both straw and reeds have a long history of use as a roofing material and have displayed good resistance to insect and fungal attack in this very adverse situation. There are different possible configurations that could be employed. Potentially the most efficient in terms of sound absorption is the end-on configuration in which the cut ends of short lengths of straw or reeds face the incident sound. Reverberation chamber measurements of the absorption coefficients of reeds arranged in this configuration revealed that for a reed length of approximately 14 cm they had a very high coefficient of absorption over most of the audio frequency range. However, the production of reed based absorbers in this configuration would require a more complex manufacturing procedure than that normally associated with the production of straw and reed products. This configuration would also require a binder or structural framework and the exposed cut ends would probably be more susceptible to insect and fungal attack. The other possible arrangement is the transverse configuration with the straw or reed lying so that the incident sound is perpendicular to their lengths. Reverberation chamber measurements of the absorption coefficients of reeds arranged in this configuration revealed that for thicknesses greater than 10 cm they had a high coefficient of absorption over most of the audio frequency range but were particularly efficient at low frequencies. Although the reed and straw configurations are porous, the cross sectional area of the pores (being very large) differs markedly from that of conventional porous absorbers. However, the absorption characteristics as measured on the larger samples indicate the potential of these or similar materials to act as very efficient sound absorbers with potential practical application. What is not understood at this time is if the primary mechanism relates to the structure of the very large pores or if micro-pores on the reed/straw surfaces are more important. Although Horoshenkov et al. [21] have presented a model for predicting the acoustical characteristics of heterogeneous material structures, this has related to pore size variation at a microscopic scale. However, the pores associated with configurations of whole reeds or straws are relatively large and the applicability of any current theoretical model to this configuration is doubtful. There is a need for theoretical studies aimed at resolving this question and perhaps leading to the development of effective novel sound absorption treatments.

Although the transverse configuration is acoustically slightly less efficient than the end-on configuration, it can be employed without a binder in a simple retaining system. The straw or reeds could also be aligned to shed water and thus the long life associated with straw and reed roofing systems might be replicated with reed absorbers located in the external environment. As part of this study, some 50 mm thick reed panels were exposed to the Liverpool climate for twelve months and were then tested to assess the effect of exposure on their acoustic performance. After this period of exposure the reed mats were found to be intact and to have suffered only a slight deterioration in the form of discolouration of those reeds on the exposed surface of the panel which also exhibited slight flaking. Measurements revealed a very slight increase in the high frequency absorption coefficients compared to those obtained with the unexposed mats. This is probably due to the flaking of the surfaces of the outer reeds resulting in a softer, more porous external surface of the panels. The results obtained from this study have shown that fibrous absorbers can be used to provide high frequency absorption and that highly heterogeneous porous materials in the form of aligned reeds can provide very good low frequency and moderate medium and high frequency sound absorption. In some circumstances a considerable amount of sound absorption is required over the entire audio frequency range and the limited area available for the application of treatment may result in the need to employ composite absorbers. The combination of fibrous materials and whole reeds offer the possibility of developing a range of very effective absorbers which act across the complete audio frequency range. Acknowledgements The results presented were developed within the IP-SME project Holiwood. This project is carried out with the financial support from the European Community within the Sixth Framework Program (NMP2-CT-2005-011799). This publication reflects the authors view. The European Community is not liable for any use that may be made of the information contained therein. The authors would like to acknowledge the contribution made by the partners of the Holiwood project with particular thanks to Montaña Jiménez Espada, UPM, Madrid, for assistance with the measurements of the composite absorber and Robert Widman, EMPA, Zurich, for facilitating the use of the reverberation chamber at EMPA. References [1] Asdrubali F. Survey on the acoustical properties of new sustainable materials for noise control. Structured session sustainable materials for noise control. [2] Asdrubali F. Green and sustainable materials for noise control in buildings. RBA-01 Acoustics and sustainable buildings. [3] Pfretzschner J, Rodriguez RM. Acoustic properties of rubber crumbs. Polym Test 1999;18:81–92. [4] Swift MJ, Bris P, Horoshenkov KV. Acoustic absorption in re-cycled rubber granulate. Appl Acoust 1998;57:203–12. [5] Horoshenkov KV, Swift MJ. The effect of consolidation on the acoustic properties of loose rubber granulates. Appl Acoust 2000;62:665–90. [6] Cook JG. Handbook of textile fibres. vol. 1 – natural fibres. Cambridge: Woodhead Publishing Limited; 1984. [7] Delany ME, Bazley EN. Acoustical properties of fibrous absorbent materials. Appl Acoust 1970;3:105–16. [8] Garai M, Pompoli F. A simple empirical model of polyester fibre materials for acoustical applications. Appl Acoust 2005;66(12):1383–98. [9] Anthony PA. The macrofungi and decay of roofs thatched with water reed Phragmites Australis. Mycol Res 1999;10:1346–52. [10] Wassilieff C. Sound absorption of wood-based materials. Appl Acoust 1996;48(4):339–56. [11] Mechel FP. Formulas of acoustics. Springer; 2002. [12] Bies DA, Hansen CH. Flow resistance information for acoustical design. Appl Acoust 1980;13:357–91. [13] Zwikker C, Kosten CW. Sound absorbing materials. Amsterdam: Elsevier; 1949. [14] Biot MA. Theory of elastic wave propagation in a fluid saturated porous soil. J Acoust Soc Am 1956;28:168–91.

D.J. Oldham et al. / Applied Acoustics 72 (2011) 350–363 [15] Biot MA. Theory of propagation of elastic waves in a fluid saturated solid. II. High frequency range. J Acoust Soc Am 1956;28(2):179–91. [16] Smith PG, Greenkorn RA. Theory of acoustical wave propagation in porous media. J Acoust Soc Am 1972;52:247–50. [17] Carman PC. Flow gases through porous media. Academic Press; 1956. [18] Attenborough K. Acoustical characteristics of porous material. Phys Rep 1992;82:179–227. [19] Stinson MR, Champoux Y. Propagation of sound and the assignment of shape factors in model porous materials having the same pore geometries. J Acoust Soc Am 1992;91(2):685–95. [20] Champoux Y, Allard JF. Dynamic tortuosity and bulk modulus in air-saturated porous media. J Appl Phys 1991;70:1975–9. [21] Horoshenkov KV, Attenborough K, Chandler-Wilde SN. Pade approximants for the acoustical properties of rigid frame porous media with pore size distribution.. J Acoust Soc Am 1998;104:1198–209.

363

[22] European Committee for Standardisation EN 10534: 1998. Determination of sound absorption coefficient and impedance in impedance tubes – part 2: transfer-function method, CEN, Brussels; 1998. [23] Horoshenkov KV, Khan A, Becot F-X, et al. Reproducibility experiments on measuring acoustical properties of rigid frame porous media (round-robin tests). J Acoust Soc Am 2007;122(1):345–53. [24] Pilon D, Panneton R, Sgard F. Behavioural criterion quantifying the effects of circumferential air gaps on porous materials in the standing wave tube. J Acoust Soc Am 2004;116:344–56. [25] European Committee for Standardisation EN 20354, Measurement of absorption in a reverberation room; 1993. [26] Ehrnrooth EML. Change in pulp fibre density with acid–chlorite delignification. J Wood Chem Technol 1984;4(1):91–109. [27] Rose KA. Guide to acoustic practice. 2nd ed. London: British Broadcasting Corporation; 1990.