Measurement series to verify the accuracy of Stora Enso Acoustic Prediction tool - SEAP

Transcription

1 PROCEEDINGS of the 22 nd International Congress on Acoustics Calculation models for timber structures (Silent Timber Build): Paper ICA Measurement series to verify the accuracy of Stora Enso Acoustic Prediction tool - SEAP Pontus Thorsson (a), Klas Hagberg (b), Andreas Golger (c) (a) Akustikverkstan AB, Sweden, pontus.thorsson@akustikverkstan.se (b) WSP Acoustics, Sweden, klas.hagberg@wspgroup.se (c) Stora Enso, Austria, andreas.golger@storaenso.com Abstract Stora Enso Acoustic Prediction tool (SEAP) is developed in order to secure a high level of accuracy in the design stage. During development a number of previously performed laboratory measurement were used in order to design the model and secure the calculated results. A third party evaluation of the draft model raised some doubts regarding the model accuracy and after some investigations it was clear that some of the laboratory measurements suffered from shortcomings and contained results that were difficult to explain, hence also creating deviations in the model. This applied to both impact sound level and airborne sound insulation. However, the model itself is designed in order to always make it possible to introduce new materials in the model as they enter the market but also to improve the accuracy as soon as new knowledge is available. Due to the uncertainties raised and in order to have the best basis prior to introduce the first version of SEAP it was decided to carry out a carefully designed laboratory measurement series and from that adjust the model to secure that the calculated results will fall within the accuracy requirements. The results from the measurement series show that it was necessary to realize some adjustments in order to reach the high requirements of the model accuracy. From the updated model new calculations were made and now the model accuracy is satisfactory. All calculations performed on various floor assemblies now fall within the accuracy limits for this stage of the model development. The model will now be further developed covering the entire delivery from Stora Enso Building Solutions, i.e. to facilitate calculations for a complete building Keywords: impact sound, airborne sound, single number, design

2 Measurement series to verify the accuracy of Stora Enso Acoustic Prediction tool - SEAP 1 Introduction Stora Enso is currently developing a new web based calculation tool for sound insulation in wooden buildings. The design tool will be available for any building using building systems by Stora Enso, e.g. residential buildings, office buildings, hotels, hospitals et.c. The tool will comprise a calculation tool but also online regulations for the countries which are the main markets of Stora Enso building solutions. Hence, a direct comparison with national regulations will be possible, adapted to where the building site is located. The prediction tool accuracy is of great importance in order to secure the final results for the clients and end users. Therefore, a third party evaluation of the draft model was carried out and this evaluation raised some doubts regarding the model accuracy. After some investigations it was clear that some of the laboratory measurements suffered from shortcomings and contained results that were difficult to explain, hence also creating deviations in the model. There were also some assumptions in the floor package models and also ceiling package models that probably underestimated their ability to increase the sound insulation. The model was too much on the safe side. This applied to both impact sound level and airborne sound insulation. However, the model itself is designed in order to always make it possible to introduce new materials in the model as they enter the market but also to improve the accuracy as soon as new knowledge is available. Due to the uncertainties that were raised during the third party evaluation for the first version of SEAP it was decided to carry out a carefully designed laboratory measurement series adapted to the system itself and from that adjust the model to secure that the calculated results will fall within the accuracy requirements. 2 Method To deliver the correct accuracy for the SEAP prediction tool laboratory measurements have been carried out. The measurement series was designed to give results with maximal use to the design of the SEAP prediction tool. No focus was thus put on testing constructions that could be applicable for e g dwellings. The prior accuracy evaluation questioned mainly the accuracy of the floating floor models for impact sound improvement and airborne sound insulation improvement, but additional measurements were included to further test the accuracy of either individual model packages in SEAP or the full prediction tool. The measurements were divided into the following; 1. Full scale measurement series following the ISO standards 2. Small scale measurement series using 1 m 2 samples to deduce parameter trends. 2

3 The full scale series comprised 16 full scale floor samples (airborne sound insulation and impact sound level) and 5 full scale wall samples. The measurement series were carried out in the accredited laboratory of Akustikverkstan AB based outside Skövde in Sweden. The small-scale measurements comprised 20 small scale samples aimed only for impact sound level. They were carried out in the same place, using a 1.0 x 1.0 meter large floating floor sample on a 4.0 x 2.5 m CLT floor. The tapping machine was placed on top of the floating floor and the vibration of the CLT plate was measured using 4 accelerometers mounted on the bottom side of the CLT plate, measuring the acceleration in the plate's normal direction. The accelerometers were mounted directly below the floating floor sample. The accelerometers were used to measure the mean surface velocity of the plate, which can be used together with the radiation factor to give a rough estimate of the radiated power from the CLT plate's bottom side. However, no attempts were done to calibrate the radiated sound power towards laboratory measurements; the vibration measurements were only used to estimate the impact sound reduction of a specific floating floor layup. A measurement for the bare CLT was also made accordingly. The impact sound reductions measured with this method are not directly comparable to fullscale laboratory measurements since it does not consider the radiation from different wave types. Furthermore it can be troublesome to include near-field effects that can be very important at low frequencies. If these drawbacks can be accepted, the small-scale set-up allows for quick and efficient testing of the impact sound behaviour of specific materials and layups. It is a very efficient method to find the impact sound reduction gradient from changing one parameter in the layup, e g screed thickness or dynamic stiffness of the elastic interlayer. 3 Results From the measurements series a number of conclusions could be drawn and model improvements were carried out. Selected examples are shown in the following sections. 3.1 Airborne sound insulation of CLT plates Airborne sound insulation of CLT plates is not a straightforward calculation task since they are orthotropic with respect to modulus of elasticity. The quotient between moduli of elasticity in the major and minor direction E major /E minor lies between 1 and 63 for standard CLT plate layups, which means that the orthotropicity can be high. A large quotient means acoustically that the critical frequencies in the major and minor directions are well separated in frequency. An isotropic plate with adapted material data can thus not approximate a CLT plate; we must model the CLT plate as orthotropic. In the measurement series it was furthermore found that the floor measurements of the bare CLT elements differ from the wall measurements. The left pane in Figure 1 compares the reduction index for 140 mm floor and wall elements, while the right pane of the same figure compares measurements for 320 mm floor and wall elements. From the figures it can easily be seen that the thicker the CLT plate, the more the airborne sound insulations differ between floor and wall elements. These differences will be further elaborated and investigated in the updated and finalized model. Apart from the critical frequencies it seems also important to calculate the limiting frequencies for bending wave behaviour. 3

4 Figure 1: Measurement results vertically vs horizontally for different CLT thicknesses In many situations a laboratory measurement is interpreted as "the truth". It is in this context, i e comparison between measurements and calculation models, important to point out that the sharp dip at 160 Hz in the 140 mm floor measurement (blue solid curve in the left pane of Figure 1) is a resonant mode of the CLT plate when it is simply supported on all four sides. This dip is thus not inherent in the CLT plate; it is an artefact coming from the combination between floor and its supporting structure. This dip has changed if, for instance, the same floor plate is supported on two sides. When comparing the sound insulation (or impact sound level) for different floor layups by themselves, i e without considering the supporting structure, such resonant modes must be disregarded from. However, they must be considered in a later stage, when a specific floor layup is inserted into a hypothetical supporting structure to be able to calculate the total resulting acoustic parameters. The sharp dip at 63 Hz for the 140 mm wall measurement is a similar resonant mode. 3.2 Impact sound level of CLT plates In the left pane of Figure 2 the measured impact sound level for the CLT 140 L5s plate is shown from two separate laboratories together with a calculation using the current SEAP model. At low and high frequencies the measurement and the calculation agrees well. In the frequency range khz the differences are larger. The reason for this is probably differences in material parameters (i e internal damping or surface softness) or it can be a result of different lay-ups in different laboratories. Furthermore the same resonance frequency at 160 Hz as could be seen in Figure 1 is visible in this figure. Measured impact sound levels and SEAP calculations for a 320 mm CLT plate are shown in the right pane of Figure 2. The overall fit of the calculation model is in this case good, but there are small differences in the low frequency. This is probably caused by the limited plate size in the laboratory measurement. In the measurement there is also a peak at 315 Hz, which probably is a resonance mode of the CLT laboratory layup. 4

5 Figure 2: Measurement results vs SEAP for impact sound 3.3 Efficiency of floating floors (floor packages) and suspended ceilings (ceiling packages) The efficiency of floating floors (floor packages) and suspended ceilings (ceiling packages) has been evaluated from both the full-scale measurements and the small-scale measurements. Figure 3 shows examples of the impact sound reduction for a lightweight floating floor made of two layers of fibre gypsum board on a 20 mm elastic board (left pane) and a heavyweight floating floor made of 60 mm concrete screed on a 45 mm elastic board (right pane). Calculation results from the model in EN :2000 is also shown in both panes. Measurements were made for more samples in the series to verify both the impact sound reduction and the airborne sound insulation improvement, but they are not shown here. Figure 3: Measurement results vs SEAP, for impact sound reduction 5

6 Both calculation models agree well above the floating floor resonance frequency for the lightweight floating floor. However at lower frequencies the SEAP model follows the measured impact sound improvement better. The measured impact sound reduction higher than zero at frequencies lower than the resonance frequency is not a measurement error; this behaviour is explained theoretically in [1] where more measurement examples showing the same behaviour can be found. The consistency between both calculation models is significantly worse when reviewing the heavyweight floating floor results (right pane of Figure 3). It is clear that the model in EN :2000 overestimates the impact sound reduction for heavyweight floating floors. This behaviour was consistently repeated throughout the measurement series. It is important to point out that the previously mentioned 160 Hz peak is seen also in these measurements, and this peak shall not exist in the calculations. The small-scale measurements were made to get empirical input to how to account for changes in the SEAP prediction tool to one constructional parameter. Figure 4 shows the material gradient of changing the concrete screed thickness. Note that the results at frequencies above 1 khz are not relevant to the full scale situation and should not be relied upon; the measurements were focused on the lower frequencies, since those are most often limiting with respect to the weighted impact sound level. In the low frequency region the difference between the screed thicknesses is clear and almost constant with frequency, at least up to 150 Hz. The results at medium frequencies must be interpreted with the small sample size in mind; a size which with this screed material has vibrational resonant modes in that frequency region. Similar constructional gradients were made for filling type and thickness, and for different elastic interlayers. Figure 4: Small-scale measurement results for impact sound reduction of different screed thicknesses 6

7 Measurements with and without a suspended ceiling were also performed for a few floating floor combinations. Figure 5 shows the efficiency for the suspended ceiling for both airborne sound insulation improvement and for impact sound reduction, when the ceiling was installed below a CLT with a lightweight floating floor and a heavyweight floating floor respectively. From the results in the figure it is clear that the acoustic interaction between the floating floor and the suspended ceiling is small; it is within the measurement uncertainty. This means that the efficiency of the floating floor and the suspended ceiling can be calculated separately. Another conclusion that can be made from Figure 5 is that a suspended ceiling has a higher efficiency for impact sound than for airborne sound in the medium to high frequency range. The results in Figure 5 must however be interpreted bearing in mind that the efficiency results are strongly affected by background noise and/or flanking transmission above 400 Hz for airborne sound insulation and above 800 Hz for impact sound level. Figure 5: Measurement results for the efficiency of a suspended ceiling 4 Accuracy examples after model adjustments In general the algorithms now fulfil the expected accuracy after minor adjustments has been made to some floor and ceiling packages, and for the CLT plate. Example evaluations after model adjustments of the SEAP prediction accuracy are shown below in Figures 6-10 for four various wall and floor build-ups. It is important to note that these layups were not included in the design work of the model. Corrections for laboratory flanking transmission or poor signal to noise ratio is not included in the SEAP prediction, which accounts for the high differences between prediction and measurement at medium to high frequencies in build-up 4 (Figure 7). 7

8 From above: Gypsum 12,5 mm service cavity 40 mm / wooden battens rigidly fixed /mineral wool CLT 100 C3s mineral wool 160 mm 7 mm plaster Figure 6: Measurement results vs SEAP updated version, for airborne sound of an external wall From above: Gypsum 2*12,5 mm service cavity 60 mm / semi-elastic metal profile / 50 mm mineral wool CLT 100 C3s service cavity 55 mm / free standing profile 50 mm / 50 mm mineral wool Gypsum 2*12,5 mm Figure 7: Measurement results vs SEAP updated version, for airborne sound of a partition wall From above: Gypsum 2*12,5 mm service cavity 60 mm / semi-elastic metal profile / 50 mm mineral wool CLT 100 C3s service cavity 55 mm / free standing profile 50 mm / 50 mm mineral wool Gypsum 2*12,5 mm Figure 8: Measurement results vs SEAP updated version, for impact sound of a floor assembly 8

9 From above: Gypsum 2*12,5 mm service cavity 60 mm / semi-elastic metal profile / 50 mm mineral wool CLT 100 C3s service cavity 55 mm / free standing profile 50 mm / 50 mm mineral wool Gypsum 2*12,5 mm Figure 9: Measurement results vs SEAP updated version, for impact sound of a floor assembly 5 Analysis The laboratory measurement series presented here has given a very good insight to the acoustic functionality of different building constructions that are currently in practical use. The measurement series has been vital in the work of adapting the SEAP prediction tool to common constructions in Stora Enso's building system. Laboratory measurements are usually performed to find good constructions, i e a specific construction is tested in the laboratory to later be reproduced in the field. This measurement series has instead been focused on understanding the acoustical efficiency of important constructional parts, which has supported the development of SEAP. The results from the measurement series has been fully implemented in the SEAP prediction tool, and in its present state the SEAP prediction tool have good accuracy for many constructions, as has been shown here. New important measurement data (from laboratory or field situations) can also be implemented when needed. 6 Discussion The model will be updated continuously which is possible due to its modular approach adapted to Stora Enso Building Solutions. The first step is finished and the accuracy using CLT elements as structural building parts, combined with various floor, ceiling and wall packages are now satisfactory. The first version of SEAP is available online during Acknowledgments The authors would like to acknowledge Stora Enso for their big effort in developing a calculation model adapted to Stora Enso Building Solutions. References [1] Gudmundsson, S.: Sound insulation improvement of floating floors. A study of parameters. PhD Thesis, Report TVBA-3017, Lund Institute of Technology, Lund, Sweden, [2] CEN, European Standard EN :2000: Building acoustics Estimation of acoustic performance of buildings from the performance of elements Part 2: Impact sound insulation between rooms