Noise attenuator

Abstract

An attenuator for ducted air flow systems such as heating, ventilation and air conditioning systems arranged to direct the airflow into the attenuator by reducing the airflow path geometry into a central channel surrounded by absorbing material to damp resonant modes to reduce noise more uniformly across the desired frequency spectrum than resistive attenuators. The attenuator has an outer skin 360, an inlet 310 and an outlet (330, figure 5c). The inlet and outlet are connected by a central channel 320 that allows air to pass through. A porous filler material 340, such as a melamine foam, is arranged around the central channel. The attenuator has a first dimension across a transverse axis, the first dimension corresponding to a half-wavelength of a primary frequency of a sound to be attenuated. The channel may be offset so that there are different amounts of porous material between the channel and the outer skin on each side of the channel.

Description

FIELD OF THE INVENTION

[0001] This invention relates to attenuators, in particular attenuators used to manage noise in air flow systems such as heating, ventilation and air conditioning (HVAC) systems. BACKGROUND

[0002] The management and mitigation of noise generated in ventilation ductwork presents a fundamental challenge to HVAC systems. It is undesirable for HVAC systems to introduce noise into the user environment. The use of fans for driving airflow through the system can lead to significant noise generation which is carried through the ductwork. In order to prevent this noise from passing into the user environment, noise mitigation is necessary.

[0003] Conventional methods for managing noise have not significantly changed for many decades. The main element used for reducing noise propagation is the so-called ‘splitter attenuator’. This kind of attenuator is typically provided as an element in the flow path with ducting feeding air into and out of it. They tend to be structures with a generally cuboid shape with several thin slabs of porous material interposed in the airpath, such that the free open area is reduced compared to the size of the airpath in the feed ducting. Figure 1 shows a typical example.

[0004] This kind of attenuator is inserted in the airflow path and fed from one duct supply. The air flow path is narrowed so that air flows through narrowed channels surrounded by porous material acting to absorb elements of the sound passing down the duct work.

[0005] The thin slabs of material (or splitters) are typically formed from a fibrous material such as mineral wool or fibreglass which is encapsulated by perforated metal sheets with an interposed glass tissue to limit fibre migration into the airpath. The resultant open area of the air flow path is often in the range 30-50%, as compared with the connecting ductwork.

[0006] The noise generated in such systems and the reduction thereof tends to focus on the typical range of audible frequencies which are encountered in buildings, typically in the range from around 63Hz up to around 8kHz.

[0007] A problem with splitter attenuators is that they exhibit an acoustic performance or (dynamic) “insertion loss” (IL) that is generally poorly suited to the environment which it is serving. There is typically pronounced insertion loss at the mid-frequencies (e.g. around 1 kHz), but much less attenuation at lower and higher frequencies.

[0008] Figure 2 gives an example of the sound level in a room for a conventional splitter attenuator (1600mm splitter attenuator). The sound level across the range of typical frequencies is shown at 110 for a typical 1600mm long splitter attenuator such as that shown in Figure 1. It can be seen that the sound level at 1 kHz is much lower than at higher and lower frequencies representing much higher insertion loss. In comparison, the insertion loss at 4kHz and 125 Hz is much less. In the UK, it is common for the target noise criterion to be defined using an internationally recognised set of spectra (so-called noise rating curve, or NR), shown in Figure 2 at 112.

[0009] Acousticians typically find themselves specifying an attenuator according to its low frequency IL. This inevitably leads to a noise spectrum in the ventilated room that is spectrally unbalanced. The excess mid-frequency IL results in much lower noise levels in that frequency range compared to that at the lower and higher ends of the spectrum. This may not appear to be problematic as long as the noise is reduced sufficiently overall. However, reducing the mid-frequency noise levels more effectively has the subjective effect of emphasizing the lower frequencies. This can lead to occupants of the ventilated room feeling that the lower frequency rumble is worse than it actually is, even though it meets the target noise levels.

[0010] A typical 1600mm long splitter attenuator such as that shown in Figure 1, meets the specification, by being lower than the NR35 curve 112. As seen in Figure 2, the curves 110 and 112 are roughly coincident at the 125Hz level so that the performance target is just met for that frequency. However, the resultant spectrum is significantly divergent by virtue of the excessive IL in the mid- and upper-frequency bands.

[0011] Conventional splitter attenuators work on a purely resistive basis which results in poor low frequency performance. This is expected given that the porous absorption used typically increases with frequency with worst performance at lower frequencies.

[0012] Another problem with splitter attenuators is that they introduce significant resistance to the airflow. This is compounded when the attenuator length is specified to satisfy the lower frequency criteria (thus overcompensating at the mid frequencies), as this results in the attenuator being more intrusive on the airflow than it would otherwise be. This causes greater losses in the airflow path. For example, the 1600mm long unit mentioned above exhibits a 70Pa pressure drop for an airflow of 400l / s. Given that both the fan intake and supply require attenuation, this resistance doubles to 140Pa. This can add significantly (e.g. 20%) to the electrical power needed to drive the airflow to overcome this pressure drop.

[0013] Authorities like the UK and EU are already considering regulations on energy efficiency which will require manufacturers to further decrease the ‘specific fan power’, i.e. the electrical power required to move 1 litre of air each second. Given that motors and impellors are already at the practical limit of energy efficiency (95% for the former) further improvements in efficiency will need to be found in ancillary items such as attenuators, to realise this reduction in energy use.

[0014] The present invention therefore sets out to address or ameliorate the problems with the existing attenuators to achieve insertion losses within specified criteria whilst reducing the overall load on the system by interposing them in the airflow. SUMMARY OF THE INVENTION

[0015] According to the present invention there is provided an attenuator comprising: an outer skin; an inlet for receiving an air supply; an outlet; a central channel passing through the attenuator connecting the inlet and outlet for the passage of air therethrough; and a porous filler arranged around the central channel, wherein said attenuator has a first dimension across a first transverse axis such that the first dimension corresponds to a halfwavelength of a primary frequency of sound to be attenuated.

[0016] With this arrangement, the selection of the combination of filler dimension and material and the dimensions of the attenuator outer skin provide a constrained space such that resonant modes can be developed which are then damped by the provided filler. By tuning the attenuator based on the materials used and dimensions, the desired frequency to be attenuated can be targeted. In this way the attenuation of specific frequencies can be targeted to provide a more balanced attenuation spectrum and avoid the issues of high attenuation at some frequencies and low attenuation at others leading to a poorer perceived noise reduction than is provided.

[0017] The central channel of the attenuator is preferably offset such that the thicknesses of the filler material between the central channel and the outer skin, along said first transverse axis, are different at either side of the central channel. In this way by having different thicknesses of filler at each side of the central channel, the attenuation and tuning can be further refined.

[0018] The attenuator may have a second dimension across a second transverse axis such that the second dimension corresponds to a half-wavelength of a secondary frequency of sound to be attenuated. The primary transverse axis can be configured to focus on a specific frequency and to an extent higher multiples (although not necessarily integer multiples due to the dependency of the speed of sound in the filler on frequency) of that frequency, which provides a more balanced attenuation spectrum. However, this may still leave some frequencies that are more or less attenuated than others and so by further tuning the attenuator using a different dimension, the spectrum can be further refined. For example, in a rectangular configuration, the vertical and horizontal transverse axes may be used. Similarly on a circular attenuator, multiple transverse axes may be tuned by modifying the thickness of the filler according to the radial position allowing for even more refinement of the spectrum.

[0019] The filler may not match the outer shape of the attenuator but may have a more complex arrangements with varying thicknesses according to the different transverse axes. For example, a circular attenuator may have a rectangular central passage way and vice versa or the central passage way may have a more complex shape according to the tuning requirements.

[0020] The filler is preferably a porous foam material. The foam material may have a flow resistivity in the range of 5,000-15,000 Pa.s / m2 and more preferably in the range 8000-12000 Pa.s / m2. A foam material with a flow resistivity of around 10,000 Pa.s / m2 is more preferable. This may be a melamine foam. This helps to provide a good balance between the damping effect of the filler material as well as the moderation of the speed of sound in the attenuator to adjust the wavelength compared to the wavelength in air.

[0021] The filler defines the walls of the central channel. By using the filler to define the walls of the channel, the need for other elements to be placed between the filler and the channel is avoided. These are often used and required in conventional attenuators, typically due to the mineral wool material used in such systems. This tends to be less structurally capable and so may not be self-supporting. It can also shed fibres which would be carried into the environment if not contained by additional elements such as sheets of tissue like material, to act as filters and containers for the mineral wool. The use of a material such as Melamine foam provides excellent sound absorption properties but is also structurally stable needing no additional support to retain it in position. Furthermore, due to its integral structure, it does not disintegrate over time and so does not shed material that would need to be captured.

[0022] The central channel preferably has a rectangular cross-section, and the portions of filler are between each of the four sides of the central channel and the outer skin of the attenuator. This arrangement is convenient in terms of manufacture; also making the outer skin rectangular provides a simple external structure with the filler material being a simple shape to be formed and inserted into the attenuator. The simpler structure also means that tuning is relatively simple yet effective.

[0023] The central channel preferably has a central portion between the ends of the channel and at least one transitional portion between the central portion and one of the inlet or outlet. The dimensions of the transitional portion vary between the dimensions of the central portion and the dimensions of the respective inlet or outlet. This transitional portion may be provided on just the inlet with the outlet having dimensions similar to the central portion of the central channel or vice versa. The dimensions of the central portion are preferably smaller than the dimensions of the inlet and / or outlet. Preferably both the inlet and outlet have dimensions different to the central portion. This transitional portion further aids the damping performance of the attenuator by transforming the duct-borne zero-order mode into higher-order modes in the attenuator which can be more easily absorbed.

[0024] The transitional portion may be formed of one of more sheet sections forming the sides of the transitional portion. The sheets can be flat or curved. Multiple sheets may be used to define the transitional portion such as by having several flat portions to form a polygonal shaped section. Equally, the entire transitional portion may be formed of a single sheet providing a liner change from one dimension to the other. Where the cross section is rectangular, the transition may be on only one of sides or may be on all four sides requiring four sheets or four polygonal or curved pieces. The use of flat sheets is often easier form a manufacturing perspective as these can be manufactured from a base sheet which is then folded to form the polygonal shape. However, curved sheets may be used and formed by pressing or other means.

[0025] Where the attenuator is round, then the transitional portion may need more complex shapes.

[0026] The transitional portion preferably forms a mouth formed from the sheets of material mentioned above. However, the transitional portion may be defined by the filler material itself, in a similar way to the central portion of the central channel. However, using solid material to define the channel aids the sound absorbing function of the attenuator and so is preferable.

[0027] The longitudinal axis of the central channel may extend through the attenuator in a direction substantially perpendicular to a face of the attenuator in which the inlet is provided. This can be convenient where the inlet and exhaust duct work is substantially linear. However, the longitudinal axis of the central channel may alternatively extend through the attenuator at an angle offset from the perpendicular to a face of the attenuator in which the inlet is provided. For example, the central channel may traverse the width of the attenuator as it passes from the inlet at the front of the attenuator to the outlet at the rear. This means that the inlet and outlet are offset which may be convenient in some installations. Having an offset channel also slightly elongates the channel and also affects the modes developed allowing for further refinement of the tuning.

[0028] The channels above are effectively straight but they may be non-linear and may for example change direction from front to back, for example the channel may traverse the width and back again along the length of the attenuator so that the inlet and outlet are aligned but the channel takes a more convoluted path.

[0029] The present invention also provides a method of constructing an attenuator comprising: an outer skin; an inlet for receiving an air supply; an outlet; a central channel passing through the attenuator connecting the inlet and outlet for the passage of air therethrough; wherein the attenuator is configured for preferentially damping a predetermined primary frequency, the method comprising: forming the outer skin with a first dimension across a first transverse axis; providing filler on at least one side of the central channel, wherein the thickness of the filler is selected so that the mean speed of sound across the first transverse axis corresponds to the first dimension being a half wavelength of the primary frequency.

[0030] The thicknesses of the filler material, if any, between the central channel and the outer skin, along said first transverse axis, are different at either side of the central channel.

[0031] The attenuator has a second dimension across a second transverse axis such that the second dimension preferably corresponds to a half-wavelength of a secondary frequency of sound to be attenuated. This dimension can then be selected to target a different frequency to the primary frequency, e.g. to provide a more balanced attenuation spectrum.

[0032] It should be noted that where the central channel has a central portion with different, typically smaller dimensions to the inlet and / or outlet, this means the air pathway changes through the attenuator. Whilst a reduction in dimensions might be expected to impact airflow, this effect is relatively insignificant as long as the open airpath is at least 40% of the supply and exhaust pathways.

[0033] The invention therefore provides highly adaptable attenuation of noise propagating in waveguides, whilst maintaining very low pressure drop. DRAWINGS

[0034] Specific embodiments of the present invention will now be described with reference to the attached drawings in which: Figure 1 shows an example of a conventional splitter attenuator; Figure 2 is a graph of sound level in a room for two attenuators compared to a reference standard; Figure 3 shows a perspective view of a damped mode attenuator; Figure 4 shows a perspective view of a variant of the damped mode attenuator; Figure 5A shows an end view of the attenuator of Figure 3; Figures 5B and 5C show cross sectional views along the lines A-A and B-B of figure 5A; Figure 6A shows an end view of the attenuator of Figure 4; Figures 6B and 6C show cross sectional views along the lines C-C and D-D of figure 6A; Figure 7 shows a perspective view of a further variant of the damped mode attenuator; Figure 8A shows an end view of the attenuator of Figure 7; Figures 8B and 8C show cross sectional views along the lines E-E and F-F of figure 8A; Figure 9 shows an example of an attenuator with a circular profile; and Figure 10 shows an example of an alternative attenuator for use in a naturally ventilated application. SPECIFIC EMBODIMENT

[0035] The following description will be based on a typical HVAC system installation in an office but HVAC systems may be applied in many other environments which will have similar constraints on noise output.

[0036] The damped mode technology described below considers the modal activity operating across the cross-section of the attenuator. The fundamental idea is to ‘transform’ the duct-borne sound energy into cut-on modes which are then damped by the lining therein. Cut-on modes are those frequencies of sound with wavelengths that are less than twice the largest dimension of the waveguide. Below the ‘cut-on’ frequency, sound in the waveguide is more difficult to attenuate. As such, there are four key variables to consider: 1. Cross sectional geometry of attenuator 2. Thickness of the porous lining 3. The reduction in cross-sectional area from the supplying duct to the attenuator 4. The specification of the porous lining

[0037] A damped mode (DM) attenuator 300 is shown schematically in Figure 3.

[0038] The DM attenuator in this example is arranged with an inlet 310 for connection to connecting ductwork measuring 471 x 302mm. The unit has a depth from front to back, i.e. in the direction of air flow, of 1050mm. The outer cross-sectional dimensions of the DM attenuator 300 are 746 x 348mm. The central passage 320 forming the air path measures 346 x 183mm giving a free area of 44.5%, i.e. the ratio of the cross-sectional area of the central passage 320 to the dimensions of supply duct and inlet 310. It will be appreciated that the above dimensions are merely exemplary and the invention is not dependent on these precise dimensions.

[0039] Figure 5A shows a view from the front of the attenuator facing the inlet. Figure 5B shows a cross section through the attenuator along line A-A in Figure 5A. Figure 5C shows a cross section through the attenuator along line B-B in Figure 5A.

[0040] As shown in Figures 3 and 5, the attenuator has a generally cuboid shape with a channel passing through it. The front inlet 310 is connected to a central channel 320 defining an airpath through attenuator, via a tapered section 315. A similar tapered section 325 between the central channel 320 and the outlet 330 at the rear of the attenuator. The rear tapered section increases in cross-section between the central channel 320 and the outlet 330. The tapered parts are formed of generally flat metal sheets 370, 380 between the respective edges of the inlet 310 and the central channel 320.

[0041] The voids between the channels 315, 320, 325 and the outer skin 360 are filled with a foam material 340. The dimensions between the outer surface of the attenuator 360 and the respective upper, lower, left and right surfaces of the central channel 320 are not all the same, in this embodiment. The spacing between the bottom of the attenuator (as viewed in figure 5A) and the lower surface of the central channel 320 is 125mm. The spacing between the right side of the attenuator and the right side surface of the central channel 320 is 125mm. The spacing between the left side of the attenuator and the left side surface of the central channel 320 is 275mm. The spacing between the top of the attenuator and the top side surface of the central channel 320 is 40mm.

[0042] Ideally, the foam material 340 is a porous filler having a flow resistivity in the range of 5,000-15,000 Pa.s / m2 and more preferably in the range 8000-12000 Pa.s / m2. A melamine foam such as Basotect ® is used as a porous filler in this example. Melamine foam has a flow resistivity of around 10,000 Pa.s / m2 which is similar to low density (10-20kg / m3) fiberglass. Flow resistivity provides a reasonable guide to the sound absorption properties of a foam but this is affected by a number of secondary factors. Foams with lower values of flow resistivity tend to result in less damping of the acoustic modes and so less attenuation overall. Low values also tend to have a lesser effect on the speed of sound in the material (see below) and so the lower frequencies are less impacted as cut-on modes cannot be established unless the dimensions of the attenuator are adjusted.

[0043] However, foams with higher flow resistivity tend to be more expensive and can also impact other properties of the foam. For example, foams with higher flow resistivity tend to have higher density and as the density increases, the sound tends to respond to the boundary of the foam more and more like a solid surface (such as the edge of the attenuator) which can impact the overall effective dimensions in which resonant modes are established restricting the potential wavelengths that can be attenuated. For example, if the foam surface appears to the sound like a solid wall, the resonant chamber is reduced to the dimensions of the central channel raising the minimum frequency and lowering the attenuation at the same time. Foams with mid-range values of flow resistivity, such as Melamine, provide a good compromise giving good damping effect, without presenting a surface that acts like a solid surface and also reducing the speed of sound to enhance the attenuation at lower frequencies for a given geometry.

[0044] Whilst, as noted above, the tapered parts 315, 325 are formed of metal sheets, they may be simply defined by the foam being shaped accordingly although, as noted below the use of metal sheet provides improved attenuation. Between the tapered section in the central channel 320 the channel is defined by the foam being shaped to define the channel profile. However, the channel may also be defined by a porous sheet or perforated metal sheet although this may undesirably affect the performance of the attenuator.

[0045] As shown in Figures 3 and 5A-5C, the dimensions taper linearly between the dimensions of the inlet and the central channel. However, the taper may be made more progressive with more rounded, bullnose profile. The bullnose profile is employed to make a smooth transition to the open airpath through the attenuator. By avoiding any sharp edges the effect on airflow resistance is reduced and helps to avoid introducing additional noise if turbulent flow is induced.

[0046] From an airflow perspective it may seem logical to allow the channel through the attenuator to be identical to that of the supplying duct. However, that regime does not lead to as significant an impedance discontinuity which means the modal activity within the attenuator is weaker. A step transition from one dimension in the supply duct to a different dimension in the attenuator passage will disrupt the airflow introducing turbulence which may actually introduce noise rather than attenuate it. It can also increase the flow resistance of the attenuator.

[0047] It is therefore actually preferable to have a transition which is more gradual to ensure a progressive transition in the airflow but also allow the resonant nodes to develop in the main channel. A linear taper aids this but, as noted above, a more progressive curved transition can improve this.

[0048] Alternatively, the taper of Figure 3 may be arranged in a series of different tapered sections 41 Oa-c, as shown in the attenuator 400 of Figures 4 and 6A-6C. These tapered sections provide a more progressive taper closer to a bullnose shape but formed of flat sheets of material, typically using the same material as the attenuator body, making manufacture much easier and reducing the costs of producing and assembling more complex curved parts.

[0049] The taper on the rear on the attenuator between the central channel 320 and outlet 330 may be configured with multiple angled sections 420a-c similarly to the front tapered channel 315 with single angled sheets like sheets 370, 380, multiple angled sections like 410a-410c, or with a more curved arrangement, such as a bullnose shape. The taper on the outlet may be dispensed with completely although step transitions can cause similar problems to those described above and so a progressive transition is more preferable.

[0050] As before, the space between the channels defined by the tapered portions and the edge of the attenuator is filled with a foam material. Although in this example, the entire space is filled with foam, voids and gaps may be left.

[0051] The attenuator insertion loss required to achieve the NR35 criterion is given in Table 1. 0052] Table 1 - Target Insertion Loss________________________________________________ Frequency (Hz) 125 250 500 1000 2000 4000 Insertion Loss (dB) 13 19 29 23 23 17

[0053] Figure 2 shows a comparison of the sound levels in the ventilated room for the design above and a typical conventional splitter attenuator 100 design such as that shown in figure 1. The comparison design is also longer, at 1600mm, than the DM attenuator design described above, which is 1050mm long from front to back. The DM attenuator is therefore 34% shorter. This can be advantageous where space constraints are an issue. These units are often installed in ceiling voids above suspended ceilings where space may be limited and so a shorter attenuator may have significant benefit.

[0054] It can be seen from Figure 2 that the resultant noise spectrum 114 of the DM attenuator is much closer to the specified NR35 curve 112 than the output 110 of the reference conventional splitter attenuator.

[0055] HVAC systems often utilise heat recovery to minimise energy loss. These systems may employ a mechanical ventilation and heat recovery (MVHR) unit which provides heat exchange between incoming fresh air and outgoing waste air, as part of the replacement of air in the building with fresh air from outside.

[0056] Many MVHR units have the supply air connection centrally located on the unit and the option to have the extract connection offset on either on one side or the other. This tends to mean the MVHR unit is wider than the connecting duct work. As a result there is generally a free space requirement on at least one side of the unit which is wider than the duct work. The design shown in Figure 3 makes use of this, by extending the width to one side relative to the duct work to make use of the redundant extract connection space. This means that the DM attenuator is not any wider than the fan it is connected to.

[0057] Table 2 - Insertion Loss for DM attenuator Frequency (Hz) 125 250 500 1000 2000 4000 Insertion Loss (dB) Target 13 19 29 23 23 17 DM Attenuator 13 24 30 32 29 19 +0 +5 +1 +9 +6 +2

[0058] The insertion loss for the DM attenuator design of Figure 3 is given in Table 2, together with the target values. The final row indicates the difference between the target insertion loss and that achieved by the DM attenuator.

[0059] In addition to satisfying the minimum octave band insertion loss values, there is only ± 9.5dB variation across the 6-octave frequency range. In other words, the maximum difference in insertion loss between the minimum attenuation at 125Hz (13dB) and the maximum at 1000Hz (32dB), i.e. a 19dB difference representing ± 9.5dB variation from a notional average across the range. This means the relative attenuation across the range is more consistent, with a much less pronounced loss in the mid-band compared to that provided by the comparison attenuator. This avoids the issue noted above where some frequencies are substantially reduced but other frequencies are still relatively strong making them more noticeable.

[0060] The cross-sectional geometry and thickness of the foam material has a significant impact on the insertion loss. In a comparison design, the dimensions are modified whilst maintaining the same relative open area at 44.5%. The comparison design has the following dimensions: Outer cross-section on attenuator: 569 x 456mm (similar area to DM attenuator) but a squarer geometry Lining thickness:                 128mm between each outer edge of the attenuator and the central channel

[0061] In the comparison design similar tapered inlets and outlets are used. Also a similar volume of foam lining material is used to the DM attenuator above.

[0062] Table 3 below provides the insertion loss figures for the DM attenuator alongside the comparison design as well as the figures for the target reference values and the difference between the comparison design and the target. The comparison is limited to four octaves.

[0063] Table 3 - Comparison of insertion loss of DM attenuator and comparison design Frequency (Hz) 125 250 500 1000 Insertion Loss (dB) Target (NR35) DM Attenuator 13 19 29 23 13 24 30 32 Comparison Design 10 30 43 41 -3 +11 +14 +18

[0064] In contrast to the DM attenuator, the target insertion loss for the comparison design at 125Hz is not satisfied. This is due to insufficient attenuator width, which limits the low frequency modal activity, as explained in more detail below. The other factor is that the lining thickness does not offer adequate low frequency absorption. Similar to the conventional splitter attenuator, the insertion loss at mid-frequencies is excessive which would lead to a negative subjective impression of the noise spectrum in the ventilated space.

[0065] This design demonstrates that despite the similar dimensions of the central passage way and foam material, the uniform lining thickness fails to provide the improved balance of insertion loss of the DM attenuator.

[0066] In a second comparison design, the dimensions of the attenuator are identical to the first comparison design above except that the cross-sectional area of the adjoining ductwork is reduced to match the attenuator airpath through the central channel. In other words, the attenuator has an open area of 100%. In a real world installation, this would result in a 55.5% reduction in air volume due to the restricted size of the ductwork and so a sound power correction has been applied to correct for this. This correction is -3.5dB which is simply calculated as 10 log™ (44.5%).

[0067] Table 4 below shows similar comparison values for the second comparison design, the DM attenuator and the reference values. Again, the difference between the second comparison design and the target is provided.

[0068] Table 4 - Comparison of insertion loss of DM attenuator and second comparison design Frequency (Hz) 125 250 500 1000 Insertion Loss (dB) Target DMT Attenuator 13 19 29 23 13 24 30 32 Comparison Design (corrected) 9 27 39 37 -4 +8 +10 +14

[0069] As with the previous comparison design, the 125Hz target is not met and the attenuation at the higher frequency values is again significantly greater resulting in poor subjective noise levels.

[0070] It has been determined experimentally that a 100% open area is not necessary in order to achieve negligible air resistance and, as noted above, enhanced performance can be achieved by having a transitional cross-section. In practice, open areas of 40% and above provide sufficient air flow with negligible effect on the airflow. Since a 45% open airpath provides the appropriate aerodynamic performance, it represents an important design feature for two reasons: 1.   A thicker lining can be accommodated for a given attenuator cross-section 2.   The area discontinuity between the supplying duct work and the central channel helps ‘transform’ the duct-borne zero-order mode into higher order modes in the attenuator which can be more readily absorbed.

[0071] Finally, a comparison is made between melamine foam such as Basotect® foam and the medium density mineral wool typically used in splitter attenuators (e.g. RW3 manufactured by Rockwool with a density of 60kg / m3). In this example, the geometry of the attenuator is identical to that of the DM attenuator described above such that the only difference is the different lining material.

[0072] Table 5 below shows similar comparison values for the third comparison design with mineral wool liner, the DM attenuator and the reference values. Again, the difference between the third comparison design and the target is provided.

[0073] Table 5 - Comparison of insertion loss of DM attenuator and third comparison design Frequency (Hz) 125 250 500 1000 Insertion Loss (dB) Target DMT Attenuator 13 19 29 23 13 24 30 32 Comparison Design 9 18 29 40 -4 -1 0 17

[0074] Table 5 shows that the coarse fibrous structure of mineral wool is not effective at damping the low frequency modes. Indeed, there is a delicate balance in providing enough damping whilst allowing the modes to develop in the first place. The excessive attenuation at 1 kHz identifies the frequency range at which the RW3 is well matched to the sound field within the attenuator. Again, the comparison attenuator fails to meet the standard at the lower frequencies but significantly overdamps the higher frequency leading to the same imbalance in the sound reduction and problematic subjective impact.

[0075] The melamine foam used in the DM attenuator provides better damping of modes within the attenuator. The low flow resistivity, together with a fine porous skeleton gives optimised acoustic resistance.

[0076] As will be apparent from the comparison examples above, the combination of the dimensions of the foam lining and the foam material itself provide optimised modal damping. This leads to attenuation which does not exhibit significant differences between adjacent octave bands. Thus the resultant noise level in the ventilated room is subjectively inconspicuous because it matches the sensitivity of the human auditory system.

[0077] The DM attenuator works by deliberately generating acoustic modes which are subsequently damped. These modes are frequencies of sound that are resonant, in the sense that they are intrinsically associated with the geometry of the attenuator. A mode is said to be ‘cut-on’ when its wavelength is less than or equal to twice the dimension across which it is operating. In order to achieve the desired attenuation spectrum, the modes must span a wide frequency range, which thus dictates the cross-sectional dimensions of the unit.

[0078] For example, effective attenuation in the 160Hz octave band would require a mode operating across a dimension of about 1070mm (which is half a wavelength at this frequency). However, the inclusion of open cell foam of the invention reduces this dimension because the speed of sound in the foam is significantly less than in air providing a lower wavelength for a given frequency than in air. The speed of sound in porous materials such as the foam, varies with frequency. The speed is closer the that in air at higher frequencies but drops at lower frequencies. In the DM attenuator design described above, the foam occupies about half of the total attenuator width. Given that the sound speed in Basotect® at 160Hz is approximately 124m / s, the damped mode at this frequency requires a minimum dimension of around 700mm. This effectively means that the average speed of sound across the width of the attenuator is around 233ms-1. At 160Hz, the wavelength across the width would be around 1,46m, so a half wavelength of 0.73m. This means cut-on modes are supported for frequencies of 160Hz for an attenuator with a width of 0.73m (which is within the 746mm size of the DM attenuator example above).

[0079] To have cut-on modes for frequencies lower than 160Hz would require an attenuator with a width of greater than 0.72m or a lower speed of sound across the width by including a higher proportion of foam. At 125Hz, the sound speed in Basotect® is approximately 113m / s. Again, assuming a 50% foam to free air ratio, this effectively means that the average speed of sound across the width of the attenuator is around 228ms-1. At 125Hz, the wavelength across the width would be around 1.82m, so a half wavelength of 0.91m. To reduce this value, for example by increasing the proportion of foam to 75% would decrease the required width to around 0.68m. However, reducing the width and increasing the foam dimensions will reduce the width of the airflow path. There is therefore a limiting factor before the airflow performance is undesirably affected.

[0080] The degree of modal damping is proportional to the amount of foam placed in that axis. For example, if no significant attenuation is required at high frequencies, one or two sides may only use a very thin foam liner, or even none at all. It will be appreciated that there is an iterative process involving both the amount of foam and the overall dimensions. The foam determines the damping and influences the frequency of a mode operating across a given axis. The cross-sectional geometry determines the maximum dimensions and how much foam can be included for a given size of air path.

[0081] The calculations above are based on a half wavelength but other harmonics can develop resonance at roughly multiples of the frequencies corresponding to the half wavelength. So if the attenuator and foam width are selected to have half wavelength for 160Hz, then frequencies around 160Hz x2, x3, x4 etc. will also be significantly attenuated. In reality, because the speed of sound in the foam increases with frequency, the peak frequencies will in practice be more than the exact multiples given above, perhaps more like 350Hz, 560Hz, etc.

[0082] Nonetheless, for a given set of dimensions, there will be a series of frequencies at which more attenuation will occur. However, given the significant damping present in the form of the lining thickness, the Q of these peaks will be low.

[0083] As noted above, the foam across the width of the attenuator provides damping but a similar effect can be achieved using the arrangement of foam and the vertical dimensions to target different frequencies. In the example above, it may be desirable to address the high frequency range, for example those above 500Hz. The vertical dimensions of the attenuator and foam can then be tailored to target these frequencies.

[0084] In the calculations above, the thickness of foam is determined but this can be arranged in uneven amounts (275mm and 125mm) either side of the central channel, as shown in the embodiment of figures 3 and 4. By employing different thicknesses of liner on each side of the air flow channel the attenuation at different frequencies can be controlled allowing control of the relative attenuation and better management of the spectrum. If a single thickness of liner was used throughout, the resultant attenuation would exhibit a more limited bandwidth as demonstrated in the results presented in Table 3.

[0085] When selecting the two respective thicknesses at either side of the width and similarly for the vertical direction, the maximum dimension is preferably no more than 15% of the wavelength (in air) of the lowest frequency to be attenuated. For example, at 125Hz, the wavelength is 2.74m which defines a maximum foam thickness of 412mm (15% of 2740mm).

[0086] In order to determine the parameters of the attenuator, it is generally preferable to start by considering the lowest frequencies. The width of the attenuator is employed as this typically has the greater dimension and is therefore able to be used to establish cut-on modes for longer wavelengths. Therefore as an initial step in designing the attenuator, the lowest frequency that is to be controlled is analysed according to the available dimensions of the attenuator and the desired attenuation. As noted above, where an attenuator with a width of 746mm is chosen, then an initial assumption of 50% of the width of the attenuator being occupied by foam leads to a primary frequency of around 160Hz.

[0087] The sound field within a duct is ‘transformed’ into higher-order modes whenever there is a discontinuity. This may simply be a change in cross-sectional area. However, the presence of a solid sound reflecting surface, such as the tapered profile of the tapered section 315, accentuates this mechanism particularly at the low frequencies.

[0088] If the porous liner exhibits a flow resistivity that is too high (which typically corresponds to a high density) then the modal frequencies are associated with a dimension that is less than the full width of the attenuator. Thus it is necessary for the porous liner to exhibit a low flow resistivity, but not to the extent that the mode is underdamped. For this reason, fibrous materials such as mineral wool do not offer optimal performance. Similarly, the inclusion of flame-retardant additives used in some foams results in excessive tortuosity. Melamine foam demonstrates a good balance of material properties by having a porosity in the range of 10,000 Pa.s / m2, good damping response and also is structurally stable and so does not shed material which is often the case with mineral wool, often requiring an additional layer to contain and fibres released. In this way, the bare Melamine foam can be used to form the channel and be in direct contact with the air flow.

[0089] Based upon the considerations outlined above, the design process for a fan serving a space which requires a certain sound spectrum can be determined. Initially, the attenuator insertion loss spectrum needs to be determined based on the ventilation system design. This is typically done by determining the sound power spectrum of the fan and the losses in the un-attenuated ventilation system. These losses occur at duct bends, splits and changes in area. The grille into the room together with any associated plenum also contribute to reducing the sound power propagating into the ventilated space. The resultant sound pressure in the room is then established based on its volume and area of sound absorption present. This is compared with a target spectrum (such as the NR35 referred to above) and the difference is the desired insertion loss that is required from the attenuator(s).

[0090] Any limits on the dimensions and cross-section of the attenuator are then established. Then the initial dimensions of the attenuator are calculated based on the low frequency attenuation required, and taking into account any other limitations.

[0091] Next, the thickness of the melamine foam needed to damp the lowest frequencies is determined. The cross-sectional dimensions are refined given the sound-speed impact of the foam. The thickness of foam on the other sides of the attenuator is then determined such that the attenuation most closely matches the required spectrum. An open area of at least 40% is preferably maintained and the reduction in area is achieved around the perimeter of the air path.

[0092] The thickness of foam is determined according to the magnitude of IL required at the lowest frequency of interest. For example, a 200mm thickness lining on both sides of an 800mm wide attenuator is taken. This represents a 50% foam thickness giving an average speed of sound of around 240ms-1 which equates to a frequency of around 150Hz for a half wavelength of 800mm. From this it can be established that cut-on will occur above 150Hz.

[0093] The foam will exhibit a maximum absorption coefficient when its thickness is equal to (or greater than) a quarter wavelength. For a 200mm thickness, this corresponds to a frequency of 155Hz. Thus in this example, we would expect very high attenuation above 150Hz because both requirements of cut-on mode and maximum foam absorption are satisfied. If the magnitude of attenuation is too high, then either the cut-on mode or foam absorption criteria can be relaxed.

[0094] The smaller dimension, typically the height (as in the example above) is used to target higher frequencies using a similar process. The iteration process then seeks to have the attenuation associated with the width ‘cross-over’ with that of the height so that target performance across the entire spectrum is achieved.

[0095] The steps in the above paragraph are then refined by iterating the process several times until the optimum insertion loss is achieved. The ultimate goal is to try to produce an IL spectrum that when applied to the generated noise in the system produces a spectrum that is balanced so that there is not disproportionate attenuation at some frequencies compared to others which would tend to lead to undesirable perceived noise levels.

[0096] As mentioned above, the change in cross-sectional area is preferably achieved using a solid element, rather than simply leaving the foam exposed. Table 6 below shows the comparative insertion loss of the tapered sections with a bullnose configuration, with one having a solid metal cover over the foam and the other bare foam forming the tapered section of the airflow channel.

[0097] Table 6 - Comparison of insertion loss for metal and foam bullnose tapered channel Frequency (Hz) 125 250 500 1000 Metal bullnose 13.4 23.0 31.3 38.5 Foam Bullnose 11.5 21.1 33.0 40.4

[0098] It can be seen that there is a reduction in low frequency insertion loss when the bullnose is simply the edge of the foam liner. This can be understood given that these low frequencies will more readily propagate through the foam liner than higher frequencies. The metal bullnose significantly reduces this effect. It is therefore preferable to use solid sheet, preferably metal, to cover the foam liner and form the tapered portion of the airflow pathway. Low density foam offers very little resistance to low frequency sound propagation. It is therefore more effective to provide a solid barrier which transforms these low frequencies into cut-on modes (assuming the right conditions).

[0099] Whilst a bullnose shape can be formed of angled sheets to replicate a gentle curve, the sheet may be formed directly into a continuously curving shape rather than the staggered flat surfaces.

[0100] Figure 7 shows a further variant of the DM attenuator described above. This design 700 includes an offset air passage. The inlet 710 is in a similar position to that of the above design but the outlet 730 is arranged in an offset position so that the central passage 720 traverses the width of the attenuator as it extends from front to back.

[0101] In this arrangement, the foam filler on each side has a more triangular or wedge shaped profile with one side tapering to a thin section 790 at the rear and the other side starting from a thin section 792 and widening towards the rear. This may be most simply achieved by mirroring one of the end caps and cutting the foam to form a slanting airpath.

[0102] The straight through design of the DM attenuator described above provides a more open airpath. This results in negligible air resistance through the attenuator but also means that the higher frequencies are less effectively attenuated. This is generally not an issue because (a) there is less fan noise generated at these frequencies and (b) losses in the ventilation system such as duct bends and grilles are significant at these higher frequencies. However, in those cases where an uplift in high frequency attenuation is required, an offset such as that shown in Figures 7 and 8A-C may be introduced between one end and the other.

[0103] In fact, this design provides improved attenuation across the frequency range, not just at high frequencies. This offset design is likely to be more expensive to produce given the need to cut the foam to the right shape. It may therefore be used in more demanding situations, where the improved performance justifies the increased cost.

[0104] As noted above, the use of melamine foam offers significant advantages over mineral wool. Mineral / glass fibre is formed from long strands of material that is much denser than a polymer. This results in a material that is generally too dense for this application, typically at least 50kg / m3. The very loose loft insulation type wool does not adequately hold its shape and so requires additional support to support it.

[0105] However other materials may be considered such as polyurethane foam, or recycled PET. Polyurethane foam exhibits a more regular structure than fibrous materials, being similar in form regardless of axis. However, the pores are typically 500pm wide, which results in low flow resistivity and a larger characteristic viscous and thermal length. As mentioned above, the addition of fire retardants increases the tortuosity, which can be thought of as being a measure of ‘structural disorder’.

[0106] For this application, it is advantageous for the pores to be small, ordered and highly interconnected. Melamine foam exhibits a high degree of cross-linking, which describes the extent to which pores are connected. This is made possible by the thin skeletal structure, which create pores 2-3 times smaller than polyurethane foam. Thus a higher flow resistivity is achieved without an increase in density.

[0107] Although the central channel is shown as square or rectangular in the examples above, the channel, and possibly the inlet or outlet may have different cross sections such as circular, hexagonal etc. Figure 9 shows an alternative construction 900 having a cylindrical construction with a circular cross section central channel 920 and a circular cross section outer surface 960 of the attenuator. The central channel 920 is offset with respect to the outer surface 960 which provides a variation in the thickness of the foam 950 with a single line of symmetry. The offset distance is determined according to the same principles described for the rectangular geometry.

[0108] The dimensions of the attenuator 900 in Figure 9 are 400mm diameter and 700mm in length. The air path is 250mm in diameter but is offset to give a minimum lining thickness of 20mm and a maximum thickness of 130mm. This attenuator is arranged so that the air path 920 diameter matched the inlet 910 connected to the connecting ductwork, such that there is no change in area (i.e. 100% open). However, the cross sectional area of the air path 920 may be less than that of the supplying duct work, so that a taper, similar to that shown in Figures 3 and 4, is provided. Having a reduced cross section i.e. <100% open is generally more effective at attenuating the noise due to the transition. The area transformation can be flat (i.e. a step change) or tapered but is preferably in the region of 60-80% open.

[0109] The outlet 930 is also similarly matched to the exhaust ductwork but may also differ.

[0110] The table below compares the performance of the DM attenuator 900 above with a conventional cylindrical comparison design attenuator. The comparison design is 1000mm long and 450mm in diameter. The dimensions are therefore greater than the DM attenuator. The lining thickness is 100mm. The lining in both cases is conventional low-density mineral wool.

[0111] Table 7 - Comparison of insertion loss for cylindrical damped mode attenuator and comparison cylindrical attenuator. Frequency ( Hz) 125 250 500 1000 2000 Insertion Loss (dB) Target IL 2 12 17 13 10 Cylindrical DM Attenuator Deviation from target 3 13 21 16 14 + 1 + 1 +4 +3 +4 Conventional attenuator Deviation from target 8 14 26 33 21 +6 +2 +9 +20 + 11

[0112] Despite the cylindrical DM attenuator being smaller in section and length, the insertion loss is much better suited to the target requirement. As can be seen from the table, whilst the overall attenuation is generally higher in the comparison design, the variation across the spectrum is much greater (±9dB) compared to the DM attenuator (±1.5dB). As explained previously, the significant extra attenuation at 1000Hz compared to 250Hz (and to a lesser extent 125Hz) leads to a worse low frequency perception to users of the ventilated space.

[0113] Although in the comparison above, melamine foam is not used, this would further enhance the performance. Furthermore, a modest reduction in the open area (e.g. 80% open), would also enhance the performance.

[0114] Whilst the arrangements described above are done so in the context or air conditioning and heating systems, the invention has application in other areas where air movement and control of noise propagation are a priority. Cross-ventilation air paths used as part of natural ventilation strategies are an example of this.

[0115] Figure 10 shows a further attenuator design for use in natural ventilation systems where there is no mechanical driving of air through a system. These are typically used between rooms often by being inserted into partitions, to allow ventilation of the rooms whilst preventing transmission of sound through between the rooms. With natural ventilation systems, because there is relatively low pressure differential driving the air flow, the airflow pathway needs to be larger to ensure adequate flow. Low air resistance is necessary because of the weak forces associated with buoyancy-driven air movement. This leads to a large opening which would make the level of sound transmission between rooms undesirably high.

[0116] Additionally, the attenuation through the cross-ventilator must be sufficiently high so as not to compromise the sound reduction of the partition / wall in which it is placed. The conventional solution is to use large conventional attenuators like the attenuator 100 shown in figure 1, which are not designed for natural flow systems and so have to be specified at a much larger size to ensure adequate airflow whilst achieving the desired attenuation. As these attenuators are often mounted in walls between rooms, if they are large they will protrude into the rooms and can be quite imposing.

[0117] A natural ventilation DM attenuator 950 such as that shown in Figure 10, provides comparable attenuation but in a much smaller unit whilst also providing a much better balanced attenuation across the spectrum of interest.

[0118] The design is simpler than the designs above with a simple cuboid shell, with a simple rectangular cross section airway 970 passing through. The ends of the airway are not tapered, so the inlet 960 extends with a constant cross-section through the attenuator to the outlet 980. The dimensions for this embodiment are 575x258mm in section and 550mm in length. The air path 970 measures 440x123mm but is offset. The lining 990 is arranged around the airway to give a lining thickness of 45mm on two sides and 90mm on the other two. The front and rear faces of the attenuator are formed using a solid material, typically metal. As this is a freestanding unit and not a ducted solution, there are no ductwork connections at the inlet of outlet and thus there is no ‘area transformation’ as such.

[0119] In the examples of Figures 3 to 8 above, the inlet and outlet are connected to the central channel by a tapered portion. However, the inlet and / or outlet may have similar dimensions to the channel such that no tapered section is needed, as in Figures 9 and 10. The tapers, if present, may have other shapes than the examples above, with anything from a step change to profiles that vary throughout the taper made of flat or curved parts. Some shapes will offer improved aerodynamic and acoustic performance but that may be balanced by increased manufacture costs dictating simpler designs which may approximate more optimum designs. It has been experimentally determined 5 that in applications where the DM attenuator is immediately adjacent to a fan unit, a tapered transition is not advantageous. In these instances, the ‘fan end’ of the attenuator is flat and presents a step-change in cross-sectional area, as compared with the outlet of the attenuator which is tapered to allow for a smooth transition to the adjoining ductwork. 10

Claims

1. An attenuator comprising:an outer skin;an inlet for receiving an air supply;an outlet;a central channel passing through the attenuator connecting the inlet and outlet for the passage of air therethrough; anda porous filler arranged around the central channel, whereinsaid attenuator has a first dimension across a first transverse axis such that the first dimension corresponds to a half-wavelength of a primary frequency of sound to be attenuated.

2. An attenuator according to claim 1 wherein the central channel is offset such that the thicknesses of the filler material between the central channel and the outer skin, along said first transverse axis, are different at either side of the central channel.

3. An attenuator according to claim 1 or 2 wherein said attenuator has a second dimension across a second transverse axis such that the second dimension corresponds to a half-wavelength of a secondary frequency of sound to be attenuated.

4. An attenuator according to according to any one of the preceding claims where the filler is a melamine foam.

5. An attenuator according to according to any one of the preceding claims where the filler defines the walls of the central channel.

6. An attenuator according to any one of the preceding claims where the central channel has a rectangular cross-section, and said portions of filler are between each of the four sides of the central channel and the outer skin of the attenuator.

7. An attenuator according to any one of the preceding claims wherein thecentral channel has a central portion and at least one transitional portion between the central portion and one of the inlet or outlet where the dimensions of the transitional portion vary between the dimensions of the central portion and the dimensions of the respective inlet or outlet.

8. An attenuator according to claim 7 where the transitional portion is formed of one of more sheet sections forming the sides of the transitional portion.

9. An attenuator according to claim 8 where the sheets are flat.

10. An attenuator according to claim 8 where the sheets are curved.

11. An attenuator according to claim 8 where the sheets are polygonal havinga plurality of flat sections.

12. An attenuator according to according to any one of the preceding claims where a longitudinal axis of the central channel extends through the attenuator in a direction substantially perpendicular to a face of the attenuator in which the inlet is provided.

13. An attenuator according to according to any one of the preceding claims where a longitudinal axis of the central channel extends through the attenuator at an angle offset from the perpendicular to a face of the attenuator in which the inlet is provided.

14. An attenuator according to claim 13 where the central channel traverses the width of the attenuator along the length of the channel.

15. A method of constructing an attenuator comprising:an outer skin;an inlet for receiving an air supply;an outlet;a central channel passing through the attenuator connecting the inlet and outlet for the passage of air therethrough; wherein the attenuator is configured for preferentially damping a predetermined primary frequency comprising:forming the outer skin with a first dimension across a first transverse axis;providing filler at either side of the central channel, wherein the thickness of the filler is selected so that the mean speed of sound across the first transverse axis corresponds to the first dimension being a half wavelength of the primary frequency.5           16. A method according to claim 15 wherein the thicknesses of the fillermaterial between the central channel and the outer skin, along said first transverse axis, are different at either side of the central channel.

17. A method according to claim 15 or 16 wherein said attenuator has a10 second dimension across a second transverse axis such that the second dimension corresponds to a half-wavelength of a secondary frequency of sound to be attenuated.15

Images