Acoustic Performance Investigation of a CLT-Based Three-Floor Building

Abstract

The acoustic performance of a CLT-based building mockup was investigated within the scope of the ADIVBois acoustic technical commission with the objective of defining wood building constructions fulfilling requirements. The CLT-based building is a three-floor construction with four rooms on each level. Measurements from junction characterization to airborne and impact sound insulations were taken. The implemented floor systems were first tested in a laboratory to evaluate their acoustic performance. Predictions based on the EN ISO 12354-1 and -2 standards were compared to building acoustic measurements. The effect of using the tapping machine and the rubber ball as impact sources was investigated both in laboratory and in the CLT-based building mockup. The effect of the apparent post between rooms is also presented with an associated simple approach to take it into account in predictions. This paper summarizes the obtained results.

1. Introduction

Wood fixes CO₂ during its growth. The new French environmental regulation RE2020 for new buildings, which is applicable from 1 January 2022, promotes the use of renewable materials to store carbon. Used as a structural element in construction, where its lifespan is around 100 years, wood is an interesting source of carbon capture. The development of wood in the construction and renovation of buildings is also recognized by the French public authorities as an effective means of contributing to the objectives of the National Low-Carbon Strategy. In addition, to achieve the carbon objectives of the Paris 2024 Olympic and Paralympic Games, aligned with the Paris Agreement, many projects are being developed with wood.

If the lightweight aspect of wood is appreciated for elevations and extensions, in multi-story buildings, the predictability of the acoustic and vibration performance of lightweight structures requires special know-how. Behavior at low frequencies and flanking transmissions must be monitored with care. Indeed, it is recommended to design such lightweight buildings with an impact sound insulation lower than the one required by the French acoustic regulation in order to improve residents’ acoustic comfort. Fortunately, research has made much progress in recent years, particularly in France, due to the momentum generated within the framework of the ADIVBois (Association pour le Développement des Immeubles à Vivre BOIS; Association for the Development of Buildings to Live WOOD), launched in 2015. The aim of ADIVBois is to help in removing technical and regulatory obstacles and to share the expertise acquired in the field of high-rise wooden constructions with project owners, project managers, and companies.

In the past years, much work has been concerned with CLT-based buildings. In order to evaluate flanking transmission, CLT junctions have to be characterized as seen, for example, in [1,2,3,4,5,6]. Contrary to concrete-based junctions, CLT junctions vary due to connectors used to secure the CLT panels (type of screws, angle brackets, hold-downs, etc.) as well as the presence of resilient elements. Jayalath [7] reviewed the recent literature on the transmissions of impact and airborne sounds, flanking transmission of timber buildings, and the state of computer prediction tools with reference to the Australian practice. The conclusion that the product variability and different construction methods in Australia enhance the requirement for detailed research into the acoustic performance of lightweight timber buildings can definitely be applied to other countries.

It has been recognized that lightweight wood floors generally do not perform well enough in terms of impact sound level, especially in the low-frequency range, leading to annoyance for building occupants [8,9]. Including one-third octave bands starting at 50 Hz for impact sound insulation, performance evaluation has become more common. A collection of well-controlled European impact sound insulation measurements of cross-laminated timber/massive wood floor constructions conducted in laboratories was presented and analyzed in [10]. The influence of the dynamic stiffness of the resilient interlayer in implemented floating systems as well the mass per unit area of the floors were found to be of importance. It should also be mentioned that within the ongoing “Sound Wood Austria” research project, the acoustic laboratory measurements of typical Austrian wooden building components were carried out in order to determine the effect of various design configurations on airborne and impact sound insulation in a systematic approach and to identify possible optimization potential [11]. Numerous configurations of wooden floors for lightweight CLT-frame buildings were also tested under laboratory conditions in Sweden [12], leading to encouraging results for solving problems of detrimental impact sound propagation in the low-frequency range in lightweight buildings made of CLT components.

In France, the technical working groups of ADIVBois have carried out work in recent years on the aspects of structure, envelope, fire safety, and acoustics. Regarding acoustic investigations, first, a laboratory test campaign on CLT floors was carried out in 2018, and then, the project for a life-sized, three-level mockup of a multi-story wooden construction named “ADIVBois Acoustic Mockup” was launched in 2019. One of the goals was to propose constructive solutions respecting the targeted acoustic performance.

Some results were presented in 2021 and 2022 [13,14,15]. This paper first presents the laboratory-measured performance on the two-floor systems implemented in a three-level wooden mockup. A comparison between the standard tapping machine and the rubber ball as impact sources is presented. Section 3 describes the three-level wooden mockup and the performed investigation in terms of measurements and prediction. The results associated with the three-level wooden building are finally presented and discussed in Section 4. First, the results relative to junction characterization, including or not the resilient layer, are shown. Then, the airborne sound insulation and impact sound insulation results are discussed. Finally, some specific cases of apparent post in the transmission path are proposed, along with a simple approach to take their effect into account in the prediction. The difference in using the tapping machine or the rubber ball is also evaluated in the three-level wooden building.

2. Laboratory Measurements

Laboratory tests were undertaken with the aim of providing designers of high-rise wooden buildings with examples of separating floors likely to comply with both French regulatory requirements in terms of construction rules as well as the comfort criteria proposed by ADIVBois for dwellings configuration. The following component performances were retained as a first step:

• R_w + C ≥ 58 dB for airborne sound insulation;
• L_n,w ≤ 52 dB and L_n,w + C_I50-2500 ≤ 52 dB for impact sound insulation.

Floor solutions with exposed wood on the underside were investigated, although this configuration is also subject to fire-safety constraints and the need to treat flanking transmissions with regard to acoustics. The base wood floor (surface area of 4.2 m × 3.6 m) is composed of two 140 mm thick CLT wood panels (4.2 m span) assembled using a tongue-and-groove-type system secured with screws spaced every 50 cm. The CLT panels are composed of five layers (20/40/20/40/20) made of spruce, corresponding to a mass per unit area of 62 kg/m². Many floor configurations integrating different types of floating systems, mass loading, and suspended ceilings were considered. The impact sound measurements were carried out in terms of L_n for the standard tapping machine and L_i,Fmax,V,T for the rubber ball as impact sources. Laboratory measurements were carried out following the ISO 10140 standard series [16]. The available laboratory acoustic test report provides more details on the investigated floor configurations (see Supplementary Materials). Furthermore, it should be noticed that the acoustic performance of different floors presented in [12] does not meet the targeted performance proposed above.

Table 1. Acoustic performance of CLT floors.

Description		Composition	Performance
System 1		1—CLT panel 140 mm	Rw + C = 66 dB
		2—Gravel 80 mm (106 kg/m2)	Ln,w = 53 dB
		3—Resilient layer 15 mm	Ln,w + CI50-2500 = 54 dB
		4—Concrete screed 60 mm	LiA,Fmax,V,T = 47 dB
System 2		1—CLT panel 140 mm	Rw + C = 48 dB Ln,w = 73 dB Ln,w + CI50-2500 = 71 dB LiA,Fmax,V,T = 64 dB
		2—Thin resilient layer 3 mm
		3—Concrete screed 50 mm
System 2a		1—CLT panel 140 mm	Rw + C ≥ 62 dB
		2—Thin resilient layer 2 mm	Ln,w = 56 dB
		3—Concrete screed 50 mm	Ln,w + CI50-2500 = 58 dB
		4—Suspended ceiling with 1 BA13 with 80 mm of glass wool	LiA,Fmax,V,T = 52 dB
System 2b		1—CLT panel 140 mm	Rw + C ≥ 65 dB
		2—Thin resilient layer 2 mm	Ln,w = 54 dB
		3—Concrete screed 50 mm	Ln,w + CI50-2500 = 55 dB
		4—Suspended ceiling with 2 BA13 with 80 mm of glass wool	LiA,Fmax,V,T = 50 dB
System 2c		1—CLT panel 140 mm	Rw + C ≥ 69 dB
		2—Thin resilient layer 2 mm	Ln,w = 51 dB
		3—Concrete screed 50 mm	Ln,w + CI50-2500 = 51 dB
		4—Suspended ceiling with 2 BA18 with 80 mm of glass wool	LiA,Fmax,V,T = 47 dB

presents some of the tested configurations and their corresponding acoustic performance. It should be noted that only System 2c was able to reach the targeted performance levels, particularly with regard to impact sound insulation, without the use of a floor covering. A flexible floor covering with a performance of ∆L_w = 18 dB on a 140 mm thick concrete floor (with, in particular, ∆L ≥ 7 dB at 50 Hz) nevertheless allows System 1 and System 2b to meet the targeted acoustic objectives.

Figure 1 illustrates the performance of floors in terms of airborne sound insulation and impact sound insulation from the tapping machine in one-third octave bands between 50 and 5000 Hz for the floor systems listed in Table 1. It can be seen that System 2’s performance is rather limited, but once a suspended ceiling was added, the performance increased substantially (System 2a to 2c).

Figure 2 presents the impact sound level for the rubber ball for the same floor systems as well as the difference between the impact sound level of the bare CLT floor and the floor system considered using either the standard tapping machine (ΔL_n) or the rubber ball (ΔL_i,Fmax,V,T). In this case, since the rubber ball measurements were performed between 50 and 630 Hz, the analysis is limited to this frequency range. In Figure 2b, it can be observed that the impact sound level difference associated with the two different impact sources is rather similar, and in general, the improvement by the floor treatment is lower for the rubber ball than for the tapping machine as the impact source.

Finally, it should be added that the carried-out tests were limited to floor solutions with a thickness of around 30 cm because the elevation constraint is particularly strong for the construction of high buildings. Solutions with concrete fill or honeycomb with granules for added mass could achieve the targeted acoustic performance, but this would require increasing the thicknesses of the floor.

3. Building Investigation

The construction of the “ADIVBois Acoustic Mockup” on the FCBA site in Bordeaux (see Figure 3) and the measurements carried out on it were intended in particular to show the in situ influence of the various components of the floors (screeds, suspended ceilings, floor finishing, etc.) and to check the suitability of the method for predicting building performance based on component performance. In addition, the study of acoustic transmission with respect to exposed posts and beams and its influence on acoustic insulation was also one of the aims of this investigation. More details on the building configuration and the complete acoustic investigations, including all measurements, are available in the project report (see Supplementary Materials).

3.1. Description

The prototype, i.e., the “ADIVBois Acoustic Mockup”, is a three-story wooden structure building comprising four rooms on each floor, including two rooms with an overall surface area of approximately 14 m² each (corresponding to volume of about 35 m³) and two rooms with an overall surface area of approximately 19.8 m² each (corresponding to volume of about 50 m³). A floor plan of the building is shown in Figure 4a; the ground level is denoted R0 and the middle level and the top level by R + 1 and R + 2, respectively.

The following building performance corresponding to a dwelling configuration with the entry level of the NF Habitat certification was selected:

• D_nT,A = D_nT,w + C ≥ 53 dB for airborne sound insulation;
• L’_nT,w ≤ 55 dB and L’_nT,w + C_I50-2500 ≤ 55 dB for impact sound insulation.

3.1.1. Walls and Floors Description

The two different floors presented previously were implemented: Floor 1 with an apparent underside (corresponding to System 1 of the previous section) and Floor 2 with a rigidly suspended ceiling (corresponding to System 2b of the previous section). Two types of vertical wall were mounted: the first is based on a 140 mm thick CLT panel, and the second corresponds to a 180 mm thick double-frame, plasterboard-based separating wall (denoted as SAD180 in the rest of the paper). The façades were standard lightweight wood-frame wall with an independent interior lining (composed of 2 BA13 plasterboards on an independent metal frame with 45 mm mineral wool).

Figure 4b illustrates this structure and localizes the different walls and floors implemented.

3.1.2. Junctions

Figure 5 presents the different junctions implemented between the different components. The junction denoted with “b” does not include a resilient layer; those with “a” do. In cross-junction LN°04, the CLT floors are connected to the vertical CLT walls using L-shaped metallic brackets spaced every 50 cm. The junctions LN°01 implement a secondary supporting beam (section of size 80 mm × 200 mm) on top of which the floor is attached. Junctions LN°02 are similar to LN°01 but without the secondary supporting beam; these junctions are not structural junctions since they are parallel to the floor span. Junction LN°05 is not symmetric; the floor is supported on a supporting beam (section of size 80 mm × 200 mm) on one side of the CLT wall and on an angle iron on the other side; furthermore, the junction LN°05b is rather a T-junction due to the presence a lightweight separating floor in the top floor. The resilient layer is a 12.5 mm thick Sylodyn NB by Getzner; it is compressed to 10 mm. Some compressed mineral wool was also incorporated for fire hazard.

Junction vibration-reduction indices were measured on the “ADIVBois Acoustic Mockup” following ISO 10848 [17]; some of these results were presented in [13] and compared to the empirical ones given in ISO 12354 [18]. Some selected results are presented in Section 4 to illustrate the effect of the resilient layer and screws at junctions.

3.1.3. Building Acoustic Performance Evaluation

The results presented in the following sections compare the acoustic performance of the building on the basis of measurements and predictions according to the ISO 12354-1 and -2 standards [18]. These predictions are based on the performance of the components measured in the laboratory (or the AcouBat software database for the lightweight SAD 180 wall) and on the vibration attenuations at the junctions measured on this mockup and the empirical ones from ISO 12354 [18]. In the tables, the results associated with prediction, as shown in parentheses, correspond to those evaluated using CLT empirical junction characteristics. However, in the results figures, only the predicted acoustic performance using the measured vibration-reduction indices on the “ADIVBois Acoustic Mockup” is shown.

Regarding the “in situ” acoustic measurements, different teams participated; the FCBA team used the in situ engineering measurement method (ISO 16283 standards [19,20]) and the other teams the survey method (ISO 10052 standard [21] and the French Acoustic Measurement Guide [22]).

Measurements and predictions were performed for different situations of the “ADIVBois Acoustic Mockup”:

• Phase 1: Bare CLT structure, with only the façade linings in place;
• Phase 2: Structure with added linings on CLT walls and added treatments on CLT floors.

The effect of leaving beams and posts apparent was also investigated.

Different floor coverings were tested. First, a simple plastic floor covering (denoted AS “RdS” in the following) with a performance of ∆L_w = 18 dB on a 140 mm thick concrete floor (with, in particular, ∆L ≥ 7 dB at 50 Hz) was used, and then, ceramic tiles were mounted using mortar. It should be noted that for simplicity, the complete floor was not covered by the floor finishing, as shown in Figure 6. These two floor coverings were also tested in the laboratory on the same CLT-based floor systems.

4. Building Acoustic Performance

4.1. Junctions

Some results concerning junctions vibration-reduction indices are given in this section for two junctions mounted with and without a resilient layer.

4.1.1. Effect of Resilient Layer on LN°04 Junction

This cross-junction was evaluated without a resilient layer (between R + 1 and R + 2) and with a resilient layer (between R0 and R + 1); in the case with the resilient layer, the junction characterization was also performed without the screws attaching the brackets to the floors (see Figure 7). The measured vibration-reduction indices are presented in Figure 7. It can be observed that the presence of the screws through the resilient layer has only a slight effect on the paths with the floor between the one-third octave bend 800 to 1600 Hz. The absence of the resilient layer is associated with a decrease of the vibration-reduction index in a low-frequency range (below 160 Hz) and then at high frequencies (above 1250 Hz). The empirical data from ISO 12354 [18] deviate from the measured data. Note that this junction is parallel to the floor span, and this could be a reason for the limited effect of the resilient layer’s presence in the connecting brackets.

4.1.2. Effect of Resilient Layer and Screws Distance on LN°05 Junction

This junction was evaluated without a resilient layer (between R + 1 and R + 2) as a T-junction and with a resilient layer (between R0 and R + 1) as a cross-junction; in the case with the resilient layer, the junction characterization was also performed without the screws and for two different screw spacings (500 and 250 mm, the last one being the recommended one also used in the absence of a resilient layer). The measured vibration-reduction indices are presented in Figure 8. For the floor–floor path, the effect of the screws and their spacing is clearly visible. For around-the-corner transmission, the paths are differentiated since the junction is not symmetric (see Figure 5; junction shown in pink frame). Once the screws are placed, the corner path on the wooden beam side is less favorable than the steel angle support. Since junctions LN°05a, integrating a resilient layer, and LN°05b, without a resilient layer, are of different types, it is rather difficult to conclude on the benefit of including a resilient layer.

4.1.3. Remarks

In general, it was observed that the empirical data for the junction vibration-reduction indices did deviate from the measured data. Indeed, this was already observed by Morandi [5], even though the measurements situation was quite different, including isolated junctions in [5] and junctions in buildings in the present work.

Furthermore, it was rather difficult to conclude about the benefit of including a resilient layer in the junctions from the results obtained in this work. It should be noted that the results presented by Morandi [5] and Schoenwald [6] demonstrated an improvement of junction vibration-reduction index in the presence of different types of resilient layers. However, the resilient layer was mounted differently: in [5,6], it was placed below the CLT panel and below the brackets, while in the present work, it was only placed on the metallic connectors.

4.2. Airborne Sound Insulation

Table 2. In situ airborne sound insulation performance.

Rooms	Prediction DnT,A|DnT,A50 (dB)		Measurement Survey DnT,A|DnT,A50 (dB)		Measurement Engineering DnT,A|DnT,A50 (dB)
S03/S04	65 (65)	57 (57)	65	49	–	–
S03/S13	60 (60)	57 (57)	58	56	60	58
S04/S14	60 (60)	57 (57)	58	57	60	58
S13/S12	65 (66)	58 (58)	–	–	66	53
S13/S14	59 (60)	53 (54)	54	50	54	51
S13/S23	66 (67)	63 (63)	62	57	65	63
S13/S24	66 (64)	62 (64)	61	60	63	62
S23/S24	57 (55)	52 (51)	–	–	55	52
S23/S22	65 (65)	57 (58)	–	–	67	58

presents the results obtained for the airborne sound insulation in terms of single-number values from measurements and the prediction. Note that the notation D_nT,A50 = D_nT,w + C_50-3150 is used when taking into account low frequencies. Figure 9 shows some of the airborne sound insulation results as a function of frequency.

The results obtained from the different measurement methods are rather consistent with the predicted ones in terms of D_nT,A, except for the horizontal transmission between rooms S13 and S14. For this horizontal transmission, the separating wall is a lightweight, double-frame, plasterboard-based wall, and it can be seen in Figure 8 that the prediction overestimates the measured performance between the one-third octave bands 80 and 315 Hz. On average, the predicted results are more in line with the measured ones following the engineering measurement method. Integrating the low-frequency adaptation terms also leads to comparable results, but the difference on average is larger. The predicted performance overestimates in general the measured one. The difference between the predicted performance using the measured vibration-reduction indices or the empirical ones is rather limited (up to 2 dB difference).

In general, the airborne sound insulation results for the “ADIVBois Acoustic Mockup” fulfill the objective of 53 dB in terms of D_nT,A based on the prediction or the measurements.

4.3. Impact Sound Insulation

Table 3. In situ impact sound insulation performance.

Rooms	Prediction L’nT,w|L’nT,w + CI50 (dB)		Measurement Survey L’nT,w|L’nT,w + CI50 (dB)		Measurement Engineering L’nT,w|L’nT,w + CI50 (dB)
S14/S04	53 (53)	54 (54)	54	56	52	55
S14/S04 (RdS)	50 (50)	51 (54)	51	54	50	54
S13/S03	53 (53)	54 (54)	54	55	–	–
S13/S03 (RdS)	50 (50)	51 (52)	–	–	49	54
S23/S13	47 (50)	51 (51)	–	–	50	52
S24/S13	50 (50)	51 (51)	53	57	52	54
S24/S13 (RdS)	47 (47)	48 (48)	51	55	47	52
S13/S14	52 (44)	51 (44)	51	48	48	46
S13/S14 (Tiles)	50 (44)	50 (44)	–	–	50	47

presents the results obtained for the impact sound insulation in terms of single-number values from measurements and the prediction. Figure 10 shows some of the impact sound insulation results as a function of frequency.

The results obtained from the different measurement methods are, again, rather consistent with the predicted ones. On average, the predicted results are more in line with the measured ones following the engineering measurement method. Some measurements using the survey methods yield results that do not fulfill the target performance when the low-frequency adaptation term is taken into account (situations identified in bold in Table 3). The predicted performance underestimates in general the measured one. The difference between the predicted performance using the measured vibration-reduction indices or the empirical ones is rather limited, except for the horizontal transmission between rooms S13 and S14 (with the separating wall being a lightweight, double-frame, plasterboard-based wall). In this case, the prediction using the empirical vibration index underestimates the floor-to-floor transmission path in the mid-frequency range (with the empirical vibration index being higher than those measured in the mid-frequency range).

4.4. Effect of an Exposed Wooden Element in a Lightweight Partition

In this section, the effect of an exposed wooden post or beam in the lightweight plasterboard separating wall (SAD180) between rooms S13/S14 as well as between rooms S23/S24 is investigated. The central wooden post of these walls has a section of 200 mm × 200 mm. Between rooms S13 and S14, there are two visible posts: one in the middle and one at the end of the partition, i.e., a linear of 2 × 2.5 m as well as a 5.5 m long beam at the top of the separating partition. Figure 11 shows rooms S14 and S24 with exposed beam and post elements; the view of rooms S13 and S23 would be identical. Figure 12 presents the enclosure type applied on the middle post in the lightweight plasterboard separating wall; similar enclosures (45 mm mineral wool with two layers of BA13 plasterboards on an independent metallic frame, i.e., the same type as the interior lining mounted on the façade) are used to encase the beams and posts.

Initially, the acoustic measurements were carried out with these posts and beams encased (results presented in Section 4.2) and then in a second test were exposed. Figure 13 shows the airborne sound insulation obtained for these configurations. The influence of the sound transmission by the visible wooden elements on the measurements performed before and after enclosing them can clearly be observed. Sound transmission through the exposed wooden structure induces a limit in the airborne sound insulation above the one-third octave band of 500 Hz. The shape of the predicted airborne sound insulation spectrum is close to that of the measurements when the wooden elements are enclosed (non-exposed).

In order to take into account the sound transmission through the wooden post in the predictions, a simple approach is proposed. An airborne sound insulation D_n,e of 62 dB for all one-third octave bands between 50 and 5000 Hz (corresponding to a performance D_n,e,w + C of 62 dB) for a length of 1 m of exposed beam or exposed post both in the emission and in the reception room for a maximum excitation surface (unfolded surface) of 200 mm × 1 m is proposed. This simple approach effectively limits the airborne sound insulation expected above the one-third octave band of 500 Hz; the predicted performance with exposed wood elements is then in line with that measured without enclosure.

Table 4. Airborne sound insulation with exposed wooden elements in light wall: Prediction and measurement.

Rooms	Prediction DnT,A (dB)	Measurement DnT,A (dB)
S13/S14—with exposed wood elements	52	52(51)
S23/S24—with exposed wood elements	53	54(53)
S13/S14—without exposed wood element	59	54(54)
S23/S24—without exposed wood element	57	55(-)

reports the associated performances in terms of the D_nT,A index. It should be noted that the effect of the exposed wood elements is overestimated by the prediction in terms of overall index (7 and 4 dB) compared to the measurements (2–3 dB and 1 dB). This is mainly due to the overestimated airborne sound insulation predicted in the low-frequency range (transmission associated with the façade path) that strongly affects the D_nT,A index in the case without exposed wood elements.

4.5. Effect on an Exposed Central Wooden Post in Vertical Transmission

The lightweight SAD180 wall between rooms S13 and S14 was then removed as well as the one between S23 and S24; thus, two large, superposed volumes were obtained (denoted as S13 + S14 and S23 + S24), with a central post of section 200 mm × 200 mm on two levels and 2.5 m high on each floor. Measurements were carried out in particular with enclosure of this post (enclosure of the same type as the lining on the façade, on the post on both levels) and without enclosure (partial enclosure corresponds to the enclosure being only on the post on one level).

It is proposed to evaluate the effect of this central column on the basis of a flanking airborne sound insulation of a specific element, D_n,f, following the given expression:

$${\mathrm{D}}_{\mathrm{n},\mathrm{f} \mathrm{post} }={\mathrm{D}}_{\mathrm{n},\mathrm{f} \mathrm{ref} \mathrm{post}}+10\log \left(\frac{{\mathrm{S}}_{\mathrm{e} \mathrm{ref} \mathrm{post}}{ \mathrm{S}}_{\mathrm{r} \mathrm{ref} \mathrm{post}}}{{\mathrm{S}}_{\mathrm{e} \mathrm{post}}{ \mathrm{S}}_{\mathrm{r} \mathrm{post}}}\right),$$

(1)

The reference post corresponds to the central post with a section of 200 mm × 200 mm and 2.5 m high in the two rooms, corresponding to that of rooms S13 + S14 and S23 + S24. This reference pole is associated with a performance of 55 dB in terms of D_n,f,w + C.

Figure 14 shows the flanking airborne sound insulation D_n,f of the reference post and its effect on the airborne sound insulation expected between S13 + S14 and S23 + S24. The presence of the exposed wood post is clear on the airborne noise insulation starting at the one-third octave band of 630 Hz. It should be noted that the partial enclosure of the post (enclosure in lower room S13 + S14 only) behaves almost like the complete enclosure of the post (enclosure in the two rooms S13 + S14 and S23 + S24).

The expected airborne sound insulation without exposed wood elements overestimates the measured one above the one-third octave band of 800 Hz; the presence of a leak or another parasite acoustic path or the limitation in terms of measurements could explain this behavior.

Table 5. Effect of exposed central wood post between superposed rooms: Prediction and measurement.

Rooms S13 + S14–S23 + S24	Prediction DnT,A (dB)	Measurement DnT,A (dB)
Without enclosure (exposed wood central post)	57	58 (57)
With partial enclosure	-	61 (60)
With complete enclosure	60	61 (60)

reports the associated performances in terms of overall D_nT,A index. It should be noted that the airborne sound insulation D_nT,A is reduced by 3 dB for the prediction as well as for the measurements due to the exposed post. In these large volumes (more than 100 m³), the airborne sound insulation in the presence of the exposed post (D_nT,A of 57 dB) remains above the regulatory threshold (D_nT,A of 53 dB). Partial enclosure is equivalent to full enclosure in terms of D_nT,A performance.

4.6. Comparison of Impact Sound Insulation between the Tapping Machine and the Rubber Ball

Figure 15 presents, respectively, the impact sound level for the standard tapping machine and for the rubber ball for the two floor systems mounted on the “ADIVBois Acoustic Mockup”; Figure 16 shows the associated impact sound improvement with respect to a bare CLT floor (as in Section 2). A clear difference can be noticed for vertical and horizontal transmission, especially for Floor 1. The difference between the impact sound level L_n in the mid and high frequency range for Floor 1 in S12–S11 and S21–S22 is due to the separating wall configuration. Once again, the impact sound improvement with respect to the bare CLT floor is rather similar for the two impact sources. As previously observed, the impact sound improvement with respect to the bare CLT floor was generally found to be lower for the rubber ball than for the tapping machine as the impact source.

Table 6. Measured impact sound performance for tapping machine and rubber ball.

Configuration	L’nT,w	L’nT,w + CI50-2500	L’iA,Fmax,V,T
Floor 1—S12–S02	51 dB	55 dB	55 dB
Floor 1—S12–S11	31 dB	36 dB	35 dB
Floor 1—S21–S11	52 dB	55 dB	54 dB
Floor 1—S21–S22	40 dB	38 dB	35 dB
Floor 2—S13–S14	48 dB	46 dB	48 dB
Floor 2—S14–S04	52 dB	55 dB	58 dB
Floor 2—S23–S14	50 dB	52 dB	56 dB
Floor 2—S24–S14	53 dB	55 dB	57 dB

shows the associated impact sound insulation in terms of single-number values quantities.

5. Conclusions

Laboratory measurements of CLT-based floors allowed selecting two types of floor that could reach the selected in situ acoustic performance. The construction and the following acoustic investigation of the “ADIVBois Acoustic Mockup” enabled a significant advance in the understanding of the acoustics of wooden constructions, in particular that of high-rise buildings made of cross-laminated wooden panels (CLT). All the data collected made it possible to verify and determine constructive solutions, integrating floors with and without a suspended ceiling.

Building performance prediction results were compared to measured ones in terms of airborne sound and impact sound insulation. The results obtained from the different measurement methods were found rather consistent with the predicted ones, and on average, the predicted results were more in line with the measured ones following the engineering measurement method. For the impact sound insulation, the predicted performance generally underestimates the measured one, which does not put the prediction on the safe side. Some impact sound insulation measurements using the survey methods yielded results that do not fulfill the target performance when integrating the low-frequency adaptation term. The difference between the predicted performance using the measured junction vibration-reduction indices or the empirical ones was found limited, probably due to the presence of linings on CLT walls and treatments on CLT floors.

It should be stressed that the obtained results do not demonstrate a significant added value of inserting a resilient layer on the floor supports on the mockup, most probably due to linings and floor treatments. However, further investigation is necessary on this specific subject.

In addition, the effect of apparent wooden structural elements, such as a post and beam located between rooms, was investigated on airborne sound transmission. A simple approach for taking an exposed wooden post into account in the performance prediction was proposed with acoustic insulation D_n,e for the visible elements integrated in a lightweight wall and with a flanking insulation D_n,f for the running elements between two rooms. These principles definitely deserve further study to explain certain observed behaviors and to extend the values to other sections of posts and beams. Nevertheless, it should be added that these exposed wooden elements (posts and beams) can pose an acoustic problem for the occupants because they constitute the main transmission of noise from 500 Hz and correspond to an identifiable propagation path that could be considered annoying.

From the laboratory and in situ measurement results, it appears that the use of the standard tapping machine is sufficient to obtain information on the impact sound performance of the investigated floors in the complete frequency range, i.e., in the low to high frequency range.

The consideration of apparent CLT walls was also evaluated in this study on the basis of the prediction model. These configurations of exposed walls deserve to be studied in more detail and especially to be evaluated in situ on the “ADIVBois Acoustic Mockup”.

This work led to the proposal of CLT-based constructive solutions that respect the targeted acoustic performances; these solutions have been shared with stakeholders in France. The targeted performance in terms of impact sound insulation is more constraining than that required by the French acoustic regulation in order to improve residents’ acoustic comfort.

Finally, it must be emphasized that the constructive solutions, in particular with respect to the exposed wood elements, must be adapted according to the constraints linked to the risk of fire, the building structure, and insurance aspects.

It is expected that the “ADIVBois Acoustic Mockup” will remain for testing other constructive solutions and other projects.