Proceedings of the Institute of Acoustics

 

 

A practical method to increase the low-frequency sound insulation of timber frame constructions

 

Michael J. Newman¹, Avinor AS, Gardermoen, Norway
Arild Brekke², Brekke & Strand Akustikk AS, Oslo, Norway
Ståle Ellingsen³, Brekke & Strand Akustikk AS, Oslo, Norway
Frode Eikeland⁴, Brekke & Strand Akustikk AS, Oslo, Norway

 

ABSTRACT

 

Timber framed houses have little or no sound insulation at low frequencies. A significant set of noise sources emit loud low frequency noise and for some of these, the noise is single frequency or tonal. Improvement of the low frequency insulation has been shown using two methods: increasing the stiffness and applying vibration neutralizers to the structure of the house. A study was carried out on a wall element to assess the viability of the method in a simplified test facility. Impedance measurements were carried out on a test section consisting of 3 studs and panel construction. An initial field implementation has been carried out on a re-purposed dwelling at the end of the runway at Bergen’s airport. Reductions of the order of 5-10dB in the C-weighted indoor noise level have been achieved. The additional attenuation achieved using vibration neutralizers only has effect when the outside noise level exceeds 87dBC. 

 

1. INTRODUCTION 

 

Sound insulation of timber frame constructions at low frequencies is limited or non-existent. This is due to the low mass, in comparison to brick, blockwork or concrete, and limited stiffness. A number of sources of environmental noise produce high levels of tonal noise at low frequencies, these include wind turbines, transformers, centrifugal fans for HVAC systems, propeller aircraft and helicopters. Other sources of high levels of broadband low frequency noise include boilers, emergency flares and fighter jets with afterburners. Previous work had been carried out on timber frame constructions to increase their sound insulation at low frequencies [1]. This indicated that increasing the stiffness of the wall gave an increase in the sound insulation at low frequencies. 

 

The intention of the work presented here was to increase the insulation for tonal noise at its fundamental and first harmonic. The increase in stiffness provided the basis to add vibration neutralizers [2-4] (tuned mass dampers) to the structure, the wooden structure was assessed as being too weak to add masses and springs at points in the structure. A FEM study was carried out on a wall model to identify suitable treatments and was followed up using a simplified test facility. Impedance measurements were carried out on a test section consisting of 3 studs and panel construction. 

 

Following this study, a field implementation was carried out on a re-purposed timber framed house at the end of the runway at Bergen’s airport. 

 

2. NOISE CONTROL MEASURES 

 

The previous work had shown that increasing the stiffness of a timber framed structure can increase the sound insulation at low frequencies. Vibration neutralizers have been implemented on many structures to reduce vibration and noise since ca 1900 [2]. These include installation on aircraft [3], pump foundations [4] and may other structures. Their addition to simple timber framed constructions is not previously reported to our knowledge. Neutralizers are useful in controlling single frequency vibration and noise. Metal structures that they are made from have loss factors of 1% or less i.e., provide attenuation over a small bandwidth. The box section used to stiffen the wooden stud provided a good “foundation” in the structure to add the neutralizers, the original stud was not structurally strong enough to support the dynamic forces exerted by the neutralizers. 

 

The aim of the study was to assess the viability of the measures to attenuate noise from helicopters, in particular the Sikorsky S92 used for transport of personnel to offshore installations. The noise spectrum produced by the S92 is dominated by its low frequency content, the blade passage frequency of the main rotor (17.6Hz) and its first harmonic (35.2Hz). 

 

A section of wall between 2 storage rooms was adapted for testing the measures. The wavelengths in air of 17.6 and 35.2Hz are so large that standardized test facilities (reverberation chambers) are not suitable to carry out these tests. The test method, given by Norén-Cosgriff et al [1] using vibration and impedance measurements, was used to assess the improvements achieved by the measures. Impact excitation and mobility analysis was used to assess whether the improvement measures would give attenuation and which of them was most likely to achieve sufficient attenuation. 

 

The chosen measures were to add a steel HUP box section to the timber frame, increasing the stiffness by a factor of 7, and to add pairs of clump masses supported by 3 cantilevers from the box section tuned to 17.2Hz and 35.2Hz. 

 

3. FIELD IMPLEMENTATION 

 

The field installation was carried out on a house built in 1957 and extended in 1977. The original construction is typical for the period with 100mm wooden studs and mineral wool insulation covered with plaster board and wood panel on the inside, building paper with air gap and weather boards on the outside. In 2004 the house received an initial round of noise control measures in the form of balanced ventilation, and a double layer of plasterboard on the outside of the studs. The windows were also replaced. The house is situated at the southern end of the runway and very close to the flight paths of the helicopters. 

 

The measures were applied to the outside of the walls by first removing the existing weather boards and plaster board. This exposed the original studs and insulation. To the studs, the HUP box section 60 x 40 x 3mm was attached with screws, an intermediate 1.8mm layer of Antifon was applied to increase damping and aid contact over the whole length of the box section. The masses were 60 x 60 x 270mm for the 17.6Hz neutralizer and 60 x 15 x 270mm for the 35.2Hz neutralizer. They were bolted in place with M20 nuts and tuned using impact excitation. Figure 1 shows the façade with neutralizers in the openings in the walls. To avoid extra damping, plastic covers were installed to hinder mineral wool ingress into the volume occupied by the neutralizers.

 

Figure 1: North-west facade of test house during installation of the treatment. 

 

4. MEASUREMENTS 

 

4.1. Measurement Method 

 

As far as possible, standardized methods were used for the pre- and post-measurements to minimize errors. The low frequency nature of the blade passage frequency from the source means that it is below the standardized frequency range for the sound insulation methods. This increases the expected uncertainty of the measurements and correction for reverberation time is not possible. 

 

4.2. Noise 

 

The sound insulation measurements were carried out following ISO 16283-3 but with the following deviations: the frequency range was extended to 10Hz – 10kHz and the indoor microphones were placed one in the central zone of the room and two 0.5m from the walls. The microphone placement was chosen as previous measurements at low frequencies gave good results. The outdoor microphone was placed in a “free field” position away from the building, ca 8m towards the airport and 4m over grade. 

 

4.3. Vibration 

 

The vibration measurements were carried out following NS8176. Though this standard is specifically aimed at surface transport, it gives a good indication of the maximum vibration levels during a pass-by or overflight experienced in a dwelling. 

 

5. MEASUREMENT RESULTS 

 

5.1. Uncertainty 

 

The results are from a single installation on a re-purposed dwelling. Flight paths for helicopters are less well defined than for fixed wing aircraft. Using offshore helicopters as sources on their normal flight paths gives a degree of variation in noise levels dependent on the following factors:

• Hight and flight path – range and angle to house 
• Speed 
• Arrival or departure 

 

The excitation of the house boundaries(walls and roof) and their final sound insulation will depend on the angle of incidence. The radar tracks given in Figure 2 show a distribution of angles, though they are mostly to the East of the house i.e., at the bedroom end of the house. 

 

Figure 2: Radar tracks of the S92s over/near the test house, departures in red, arrivals in blue. 

 

Timing between the inside and the outside measurements was not absolute as the external measurements were carried out using a stand-alone sound level meter manually synchronized at the start of the measurements. 

 

At low frequencies 1/3 octave bands do not cover a large frequency range (16Hz 1/3 octave has a 3.7 Hz bandwidth). To achieve a BT (Bandwidth x Time) product of 100 a measurement of 27s is required. Typical flyovers where the analysis was performed on the top 10dB of the C-weighted level lasted about 10s for arrivals and 15s for departures. Though these were averaged over several movements, the low BT product sets a limit as to the available uncertainty. Uncertainty at low frequencies is not expected to be better than +/- 3dB even for the two 1/3 octave bands added together to cover the blade passage frequency when doppler shifted for flying towards and away from the house and the uncertainty for C-weighted levels are not expected to be better than +/-2dB. 

 

The outdoor microphone was placed 4m over the ground. Cancellation, due to a strong ground reflection, can clearly be seen in some individual flyovers in the 40Hz 1/3 octave band, see the typical example in Figure 3, with a sharp dip in external noise level at the 21st second of the flyover.

 

Figure 3: Sound levels in 1/3 octave band 40Hz during flyover. 

 

The vibration neutralizers can be seen to provide non-linear attenuation – see Figure 4. Comparison of the orange (inside) and grey lines (expected inside level from; outside level – average sound insulation 18.2dB) shows that the expected level (with linear attenuation) sees a further 5dB of attenuation over and above the linear attenuation. They appear to hold the sound level just below 70dBC inside. This level persists a few seconds after the inside level would be expected to decay. This may be due to energy storage in the structure as well as the difficulties in synchronizing the measurements inside and out. This compression is not seen on all flyovers for all rooms even though all show improvement due to the installation of the vibration neutralizers and stiffening of the studs. The initiation of the additional attenuation is consistent with an outdoor sound level of 87dBC. 

 

Figure 4: C-weighted sound levels inside bedroom 1 (orange) and outside (blue) vs time during flyover post-installation measurements. The grey line is an estimated inside level based on the outside noise level – the average sound insulation. 

 

The factors listed above render simple single value evaluation of the results difficult. It should be remembered that the results are given for one test house. The results are an ensemble average of mostly departures. Variation in sound levels and BT product at low frequencies limit accuracy.

 

5.2. Sound Insulation 

 

Standardized measurements are only valid down to 50Hz for sound insulation. This is due to the difficulties in measuring the reverberation time/lack of room modes below 50Hz to form a statistical basis for the reverberation time; this is well above the blade passage frequency and its first harmonic. The low frequency difference spectra from 10 to 125Hz and the C-weighted difference shown in Figure 5 for post-treatment measurements. The spread of results shows the uncertainty in the low frequency measurements. The lounge was the only room where only the stiffening of studs and installation of neutralizers was applied – the room already had sealed unit glass in the windows containing laminated glass.

 

 

Figure 5: Difference values for flights and average for lounge on the main floor – post-treatment.

 

6. IMPROVEMENTS 

 

6.1. Vibration 

Measurements carried out following NS8176 show significant reductions in the statistical maximum value of the floor vibration. The results from the vibration measurements show a clear improvement with the vibration reduced to about 1/3 of the initial level on the floor of the lounge on the main floor, see Table 1. The highest reduction was 62% for the lounge during arrivals. The vibration levels are just above the limits in NS8176 for land-based transport for new dwellings (Class C 0.3 mm/s). 

 

It should be noted that no vibration treatment was implemented on the floor, the measures were installed only in the ceiling and walls i.e., limiting ingress of the low frequency sound that excited the floor. The values for the lounge ceiling show an increase for departures but a decrease for arrivals, both post treatment values being around 0.6mm/s. 

 

Table 1: Vibration velocity following NS 8176, vw,95 in mm/s 

 

 

6.2. Sound Insulation 

 

The results for standardized sound insulation measurements in the frequency range from 50Hz and upwards show little or no improvement as expected.  

 

The most consistent values of sound insulation are those given in Table 2. The A-weighted difference values ranged from 27 to 35dB for the pre-treatment measurements. The post-treatment values are all above 32dB. The improvement in the A-weighted level difference is mainly due to the installation of laminated glass windows, where sealed unit glass with different glass thicknesses were present before treatment. 

 

Table 2: Difference values for A- and C-weighted sound insulation with S92 as sound source 

 

 

The effectiveness of the stiffening and neutralizer treatment is best assessed by looking at the C weighted difference values. Similar to the A-weighted difference values the C-weighted difference values for the pre-treatment measurements vary significantly 7 – 13dB for the main floor (6dB range), for the post-treatment measurements the values range from 16 – 19dB i.e., smaller variation (3dB range). The largest improvement being for the kitchen with almost 11dB, typical improvement was 5-6dB in the C-weighted level difference for the main floor. For individual flights the compression effect, as shown in Figure 4, gives more attenuation of the highest sound levels experienced during a flyover, though the variation between different flyovers makes the assessment of this additional attenuation more complex. 

 

The improvement in individual 1/3 octave band (or pairs of 1/3 octave bands) have a higher uncertainty than for C-weighted values. Figure 6 shows the improvements achieved for the 1/3 octave bands from 10 to 125Hz and the C-weighted difference. In general, the values for the 16 and 20Hz 1/3 octave bands are improved such that they exceed 10dB. The results for the 31.5 and 40Hz 1/3 octave bands show little or no improvement. The most likely factor for the short fall is the cancelation caused by the ground reflection shown in Figure 3. 

 

Figure 6: Difference spectra for pre and post treatment, and improvement (negative values are poorer) for bedroom 2 on the main floor

 

7. CONCLUSIONS 

 

The results show that the low frequency noise control measures give measurable improvements in the sound and vibration levels in the test house during helicopter flyovers. Vibration levels produced by the low frequency sound exciting the floors have been significantly reduced by the vibration neutralizers and stiffening measures. The vibration levels are reduced to 1/3 of the pre-treatment levels even though the treatment is not directly applied to the floor. 

 

The A-weighted sound levels are improved but not significantly compared to the uncertainty in the measurements. The C-weighted sound levels are improved with the average improvement for individual room of the test house in the range 5-10dB. After implementation of the stiffening measures and vibration neutralizers the difference between the inside and outside C-weighted sound levels was measured for the test house to be 16-19dB for the timber frame section of the house. The pre-treatment measurements showed differences in the range 8-13dB. This is a significant improvement. The sound insulation achieved also exhibits non-linear attenuation, in that the internal sound level follows the external noise level to circa 87dBC and then reaches a ceiling above which it does not rise. This indicates that the vibration neutralizers are effective for reduction of high sound levels. Below this level the attenuation is governed by the increase in stiffness achieved by the box section steel tubes. 

 

Up to 10dB of attenuation, in the 1/3 octave bands where the neutralizers were tuned to, has been achieved. In the pre-treatment measurements the 1/3 octave differences varied from 0-6dB in the sum of the 16 & 20Hz bands, the post-treatment differences were all increased to around 10dB. Some improvements were shown for the 31.5 & 40Hz bands, but the results were inconsistent. 

 

Feedback from those using the test house is that the treatment makes a large improvement. The floor vibration is commented as a reduction from a feeling of “frothing” underfoot to just noticeable. 

 

8. ACKNOWLEDGEMENTS 

 

The Authors wish to thank Avinor for funding the project as part of their ongoing work to provide a good external environment at airports around Norway. The Authors also wish to thank SWECO who carried out the pre- and post-measurements. 

 

9. REFERENCES 

 

1. Countermeasures against noise and vibrations in lightweight wooden buildings caused by outdoor sources with strong low frequency components by Karin Norén-Cosgriff, Finn Løvholt, Arild Brekke, Christian Madshus and Halvard Høilund-Kaupang, Noise Control Engr. J. 64 (6), No vember-December 2016, pp. 737-752 

2. Passive Vibration Control by D.J. Mead, pub. Wiley 1999 

3. MD-80 aft cabin noise control, case history. By M.A. Lang, D.R. Lork, D.N. May & M.A. Simp son 1992 NASA/SAE/DLR 4th aircraft interior noise workshop pp. 13-33 

4. Vibration Neutralisers for Controlling Pump Noise by M.J. Newman & K. Nordmark Inter-Noise 1996 pp. 1645-48 

5. ISO 16283-3 Acoustics — Field measurement of sound insulation in buildings and of building elements — Part 3: Façade sound insulation 

6. NS 8176 Vibration and shock, Measurement of vibration in buildings from land-based transport, vibration classification and guidance to evaluation of effects on human beings

 

---

¹ michael.james.newman@avinor.no 

² ab@brekkestrand.no 

³ ste@brekkestrand.no 

⁴ fei@brekkestrand.no