A FEM-based planning tool for the vibro-acoustic design of wooden floors at low frequencies

Transcription

1 A FEM-based planning tool for the vibro-acoustic design of wooden floors at low frequencies Mathias KOHRMANN 1 ; Martin BUCHSCHMID 2 ; Ulrich SCHANDA 3 ; Gerhard MÜLLER 4 1 Hochschule Rosenheim, Germany 2 Technische Universität München, Germany 3 Hochschule Rosenheim, Germany 4 Technische Universität München, Germany ABSTRACT Within the scope of the recently finished cooperate research project VibWood at Technical University Munich and Rosenheim University of Applied Sciences, fully parametrized numerical models for wooden floor constructions have been developed and calibrated by experimental modal analyses. Based on a vast parametric study including floating floors and suspended ceilings a database of narrowband sound power levels in the frequency range up to 125 Hz for a wide range of floor dimensions has been set up. A procedure will be presented on how to process the numerical data and derive single number values that allow a comparison to the standardized rating of impact sound insulation according to DIN EN ISO The derived data is implemented into a graphical user interface (GUI). This GUI is available for free. The database is open for additional geometries and will be enhanced by different constructions in the near future. Keywords: Vibro-acoustic, Impact noise, Design guideline I-INCE Classification of Subjects Number(s): 51.3, 72.4, 74.1, 75.3, INTRODUCTION Timber constructions in multi-storey buildings for industrial as well as for residential use have been rapidly growing its market share in recent years. This can be explained on the one hand by reasons of profitability. Wood is light in weight; therefore a high degree of prefabrication is possible and also leads to less transportation and storage costs. On top of that adding another storey to an existing building is often only possible using timber constructions. On the other hand regarding the necessity to develop sustainable construction concepts timber constructions have advantages as wood is a renewable raw material. The serviceability linked with vibrations as well as the proof for footfall noise insulation are key issues for such light-weighted structures, especially for timber floors. In particular in the frequency range below 125 Hz timber structures are prone to pedestrian induced vibrations, footfall noise as well as re-radiated sound caused by induced vibrations. During the last decades the human perception of floor vibrations has been subject to a large number of researches. Several design guidelines have been proposed starting from the limitation of deflections over subjective assessment-based methods towards dynamic response-based methods, as in actual European standardizations [1]. Most of them mainly focus on the limitation of the fundamental natural frequency achieved by simplified calculation models. These often lead to a conservative prediction [2]. In practice a modified method is widely used which has been developed during a research project at Technical University Munich (TUM) [3]. In the absence of a practically applicable design guideline, simplified models are used for the prediction of the footfall noise insulation of wooden floor constructions [4]. This leads to comfort problems that arise, even if the requirements according to actual standardizations are met [6]. The same holds for concrete floors. Booming of the floating floor was identified as a possible reason, for 1 mathias.kohrmann@fh-rosenheim.de 2 martin.buchschmid@tum.de 3 ulrich.schanda@fh-rosenheim.de 4 gerhard.mueller@tum.de 3743

2 which Burkhard [7] proposed an evaluation model. In addition, in case of wooden floors a discrepancy is observable between subjective perception and measurement based proof, which is caused by the difference between the excitations of a walking pedestrian and the standard tapping machine. Rabold [6] compared alternative design approaches and derived target values for the design of wooden floors. In summary there is a requirement for more detailed models to predict the vibro-acoustical behavior of wooden floor constructions. Since measurement based investigations are time consuming and cost intensive a possible alternative is the development of numerical models. In recent years several research has been carried out on that topic [8] [9] [10]. The scope of the recently finished research project VibWood was the development of parametrical Finite-Element (FE) models of wooden floor constructions for the simulation of the vibro-acoustical behavior in the frequency range up to 125 Hz and the development of design guidelines [11]. 2. Research project VibWood The project was carried by a cooperation between the TUM and the University of Applied Sciences Rosenheim (HSRo). It was intended to investigate representative floor constructions to cover the main common wooden floor types. Therefore lumber joist (LJ) and cross-laminated-timber (CLT) floors in varying construction states with floating floor (FF) and suspended ceiling (SC) were examined. For every floor construction a fully parametrized numerical model has been developed at the TUM (see fig. 1). The models have been calibrated and updated by means of modal parameters (Eigen frequency, modal damping and mode shapes) derived by measurement based investigations [12]. Therefore experimental and operational modal analyses have been carried out in the floor testing facility of the Labor für Schallmesstechik (LaSM) at the HSRo (see fig. 2) [13]. Figure 1 FE model of a CLT floor with FF and SC Based on the calibrated, numerical models extensive parametric studies have been performed extending dimensions of the testing facility and thus extrapolating over a wide range of practical relevant floor dimensions. Modal and harmonic frequency response analyses have been carried out in the frequency range up to 125 Hz aiming at evaluation of the Eigen frequencies as well as the frequency dependent radiated sound power of the individual floor constructions. A simulation of the total frequency range, which is relevant in building constructions, was not possible, since the occurring small wavelengths cannot be modeled by the applied FE method. The adjacent receiving room was not part of the models in order to reduce the computational effort. Thus the evaluations aimed at the prediction of the radiated sound power. It was approximated according to the theory of a plane radiator as sum of plain waves [14] out of the simulated structure borne velocity pattern at the floors lower surface. Therefore an acoustical evaluation model based on integral transform methods (ITM) was developed, which analyses the simulated vibration response in the wave number domain [15]. A high precision compared to experimental data could have be achieved. 3744

3 Figure 2 Example of a measured frequency response function at the driving point with visualizations of the corresponding Eigen modes. Thus a data base for every parameter combination and construction state was derived containing the individual Eigen frequencies as well as the resulting radiated sound power. The later was stored as small band transfer functions related to the induced excitation. The derived data is accessible for engineering purposes via a graphical user interface. 3. Graphical user interface The graphical user interface developed within pursuing research at HSRo combines multiple evaluation possibilities. An example is shown in figure 3. Since the program is designed in German language the following section will provide hints on the translation. The floor constructions of interest (Deckensystem) as well as its dimensions (Länge, Breite) can be chosen in the input section (Eingangsdaten). Further an excitation type (Lastspektren) can be chosen for an acoustical evaluation as well as a specific weighting function (Bewertungsfilter). Thus the stored transfer functions are multiplied by the load spectra of a walking pedestrian or a standard tapping machine to derive the resulting radiated sound power. Up to three different floor constructions can be displayed (different color) and compared for both types of excitation. The acoustical evaluation can be done in three different ways with varying degree of detail. The standard configuration shows the third octave spectra of the sound power level from 31.5 to 100 Hz. For more detailed considerations the small band results accompanied by the individual radiation efficiency (Abstrahlgrad) can be opened in a Pop-Up window. A fast comparison of different constructions can be done by single values, also a classification according to the predicted standard footfall noise level can be done. The derivation of the single value is subject of the next section. As an additional functionality the statutory proof of the serviceability linked to vibrations (Gebrauchstauglichkeitsnachweis) can be provided for the chosen floor. This can be done according to EUROCODE 5 or optional by the modified method developed by Hamm and Richter [3]. An advantage compared to already existing tools consists of the possibility to compare the calculated to the simulated results. Thus possible oversizing caused by the simplified calculation approach can be identified and avoided. 3745

4 Figure 3 Graphical user interface; a comparison of the radiated sound power (Schallleistungspegel) of three different floor constructions for excitations by a pedestrian (Geher) and by the standard tapping machine (Hammerwerk) 4. Single values for the evaluation at low frequencies A direct calculation of the standard footfall noise level L n,w according to [5] is not possible using the simulated data, due to the insufficient simulated frequency range. Thus a coefficient derived from the simulated data, must be developed, which correlates with the single value L n,w. Therefore in a first step the radiated sound power is evaluated for an excitation with a standard tapping machine. The resulting small band spectra are subsequently transformed into third octave bands. The sound power level can then be transferred into a sound pressure level under simplified assumptions of a diffuse sound field according to equation (1). L p, i, diffus 2 4m LW, i 10log10 (1) A For reasons of comparability of the results the equivalent sound absorption area A eq of 10 m 2 (=A 0 ) is chosen according to DIN EN ISO To consider the psycho-acoustical perception the results are further processed by applying the perception threshold according to DIN [16]. Figure 4 shows an example of a third octave sound pressure level with superimposed perception threshold. The sum level of the values above this threshold serves as the desired coefficient. An investigation of the simulated results showed, that the sum level from 50 to 100 Hz (L sum, ) gives quite satisfying results. The third octave bands below 50 Hz have only a minor impact on the result. In addition, this frequency range is covered by the usual measurement frequency range of impact sound insulation. As a validation the developed coefficient is compared to the loudness computed out of the same set of data according to the method proposed by E. Zwicker [17]. As shown in Figure 5 a good correlation can be achieved. eq 3746

5 Figure 4 Example of a simulated sound pressure level of a LJ floor with FF and SC (blue); superimposed: the perception threshold (orange) and the values above the threshold (red) Figure 5 Correlation of the coefficient L sum, compared to the loudness by Zwicker for LJ and CLT floors exited by the standard tapping machine 3747

6 A similar evaluation can be performed for an excitation by a force spectrum of a walking pedestrian [18]. In this case the lower frequency bands do have a major impact. Thus the coefficient for a pedestrian excitation is formed by the sum level of the third octave bands from 25 to 100 Hz. Figure 6 shows again the correlation to the loudness. A force spectrum of a pedestrian of 75 kg with a footstep frequency of 1.6 Hz was applied. Again a good correlation can be observed. Especially the floors with an additional screed (FF) show lower sum levels respect. loudness. Figure 6 Correlation of the coefficient L sum, compared to the loudness by Zwicker for LJ and CLT floors with pedestrian load The coefficients out of the simulated data cannot directly be compared to the impact sound level as mentioned before. Instead a comparison based on measurement data is performed as shown in figure 7. It shows the database for varying wooden floor constructions archived by the IFT Rosenheim, out of which the construction catalogue for [19] was derived. Here the coefficient L sum, is compared to the standard footfall noise level L n,w complemented by the spectrum adaptation value C I, A good linear correlation is observable. The values differ only by a fixed value of 12.5 db with a standard deviation of ±1.8 db. Thus the estimated L n,w + C I, can easily be extrapolated from the coefficient L sum, For a maximum footfall noise level of 53 db, as proposed by [6] for a sufficient comfort level, a floor construction meets the requirements for a L sum, lower than 43 db. A comparison between an excitation by a standard tapping machine and a pedestrian load is shown in figure 8. No correlation can be observed, but clearly the different construction states form clusters. Floors with FF in general have lower coefficients for excitations by a pedestrian, even if the coefficient linked to the standard tapping machine excitation vary. This behavior might be caused by the simulated FF, which was chosen to have a high mass-spring-mass resonance frequency at about 100 Hz according to the goal of the Vibwood project. Thus the typical booming did not appear and the FF was acting only as additional mass in the frequency range below 100 Hz. At the same time this decreases the effectiveness of the FF with respect to impact sound insulation. Further research is intended, which includes the simulation of varying mass-spring-mass resonance frequency to complement the database. 3748

7 Figure 7 Correlation between the coefficient L sum, and the impact sound level with spectrum adjustment value L n,w + C I, for measurement data of different wooden floor constructions (IJ: I-joist floor, HBG: hollow-box-girder); superimposed: a linear trend line with 95% confidence bounds Figure 8 Correlation between the coefficient L sum, (tapping machine) and L sum, (pedestrian) for LJ and CLT floors 3749

8 5. CONCLUSIONS The presented planning software will provide the practical engineer with the advantages of a numerical simulation without the cost and time intensive calculation effort. The disadvantage of the limited frequency range could be compensated by an adequately chosen coefficient which allows an interpolation for the estimated standard footfall noise level. The combination with the possibility to provide the proof of serviceability linked with vibrations contributes an additional value especially for the prevention of oversizing the constructions. Up to now the database is limited to only floor constructions of a few fabricants, but the program can be complemented by additional systems. ACKNOWLEDGEMENTS The IGF project N des ivth (Internationaler Verein für technische Holzfragen e.v.) was supported by the AIF (Allianz Industrie Forschung). Technical support was provided by Müller BBM VAS. REFERENCES 1. Weckendorf, J., Toratti, T., Smith, I., Tannert, T.: Vibration serviceability performance of timber floors. Eur. J. Wood Prod., Springer-Verlag, Berlin, Heidelberg, DIN EN : Eurocode 5: Bemessung und Konstruktion von Holzbauten; Teil 1-1: All- gemeines - Allgemeine Regeln und Regeln für den Hochbau. Deutsches Institut für Normung e.v., Beuth Verlag 3. Winter, S., Hamm, P., Richter, A.: Schwingungstechnische Optimierung von Holz- und Holz-Beton-Verbunddecken: Entwicklung von Bemessungs- und Konstruktionsregeln auf der Grundlage eines kalibrierten FEM-Modells und unter Berücksichtigung des Dissipations-Potentials unterschiedlicher Deckenaufbauten. TU München, Forschungsbericht, DIN 4109: Schallschutz im Hochbau - Anforderungen und Nachweise. Deutsches Institut für Normung e.v., Beuth Verlag 5. DIN EN ISO 717-2: Akustik Bewertung der Schalldämmung in Gebäuden und von Bauteilen Teil 2: Trittschalldämmung. Deutsches Institut für Normung e.v., Beuth Verlag 6. Rabold, A.: Trittschalldämmung richtig bewerten. 1. Internationale Schall- und Akustiktage, Burkhart,C.: Tieffrequenter Trittschall Messergebnisse, Beurteilung. Tagungsband DAGA, Rabold, A.: Anwendung der Finite Element Methode auf die Trittschallberechnung. Technische Universität München, Dissertation, Flode, O.: Vibrations in lightweight structures - efficiency and reduction of numerical models. Lund Universität, Schweden, Dissertation, Churchill, C., Hopkins, C., Krajci, L.: Modelling airborne sound transmission across a hybrid heavyweight-lightweight floor using Statistical Energy Analysis. Forum Acusticum, Kohrmann, M., Völtl R., Müller G., Schanda U., Buchschmid M.: Planungshilfen zur schall- und schwingungstechnischen Beschreibung von Holzdecken und zur Bewertung und Dimensionierung von angepassten Schwingungsschutzsystemen. TU München und HS Rosenheim, Research report, Kohrmann, M., Buchschmid, M., Greim, A., Müller, G., Schanda, U.: Vibroacoustic Characteristics of light-weighted Slabs - Part 1: Aspects of Numerical Modeling, Model Updating and Parametric Stu- dies using the Buckingham π-theorem. Proceedings AIA-DAGA, Meran, Völtl, R., Kohrmann, M., Buchschmid, M., Schanda, U., Müller, G.: Messung und Berechnung der Schwingungen von Holzdecken im Prüfstand. Tagungsband DAGA, Darmstadt, Cremer, L., Heckl, M., Petersson, B. A. T.: Structure-Borne Sound - Structural Vibrations and Sound Radiation at Audio Frequencies. Springer-Verlag, Berlin, Heidelberg, Kohrmann, M., Buchschmid, M., Müller, G., Völtl, R., Schanda, U.: Numerical models for the prediction of vibro-acoustical characteristics of light- weighted ceilings, Proceedings Internoise, Innsbruck, DIN 45680: Entwurf. Messung und Beurteilung tieffrequenter Geräuschimmissionen. Deutsches Institut für Normung e.v., Beuth Verlag 17. DIN 45631: Berechnung des Lautstärkepegels und der Lautheit aus dem Geräuschspektrum - Verfahren nach E. Zwicker. Deutsches Institut für Normung e.v., Beuth 3750

9 Verlag 18. Schanda, U., Tröbs, H.-M., Völtl, R., Becker, P.: Semi-empirical model of the impact force of a walking person in the time domain and generated impact sound spectra, Proceedings Internoise, Innsbruck, DIN : Entwurf. Schallschutz im Hochbau - Teil 33: Eingangsdaten für die rechnerischen Nachweise des Schallschutzes (Bauteilkatalog) Holz-, Leicht- und Trockenbau, flankierende Bauteile. Deutsches Institut für Normung e.v., Beuth Verlag 3751