Proceedings of the Institute of Acoustics

 

 

Characterisation testing of floating floor systems under impact conditions

 

Adam Fox¹, CEng MIMechE AMIOA, Mason U.K. Ltd, Farnham, United Kingdom

 

ABSTRACT

 

Characterising the nature of vibration produced by impact events from sporting activities remains a challenge. The variables of the impactor, isolation measures and response of the supporting structure inherently make this and prediction of resultant noise levels a significant challenge.

 

This paper presents research into the performance of floating floor systems. Under several excitation events such as weight drops and treadmill activity, the behaviour of the floating floor was characterised through generation of a validated finite element model. The insertion loss of the system was confirmed through extensive testing for a wide range of typical gymnasium activities. The effect of floor dampers and use of covering layers was also investigated.

 

This paper presents these results which provide a validated understanding of how varying the nature of the impact generating activity affects floating floor insertion loss and how the floating slab responds dynamically to impact. This information provides data suitable for extrapolation to other installations, which is highly sought by industry to predict the implication of gymnasia installations.

 

1. INTRODUCTION

 

1. Typical gymnasium activities can generate significant impact forces, which can cause unwanted noise elsewhere in a building. This is a common problem, especially for residential developments with gyms adjacent to living areas. It is also common to see gyms installed in existing buildings which were not designed for the purpose.

2. Most gym activities are formed around energetic activities such as running, cycling, the dropping of weights. Energy from these and other activities impart impulsive energy into the floor which, if not controlled, can cause a vibration which reradiates as noise.

3. There are tried and tested methods of controlling impact noise on the market, but all have implications to the end client and selecting the most efficient solution can be difficult because of inherent complexities; the performance is dependent on not only the activity but the specific structural response to that activity. The nature of the existing structure is a very significant variable, rendering most generic test data ineffective.

4. Mason Industries have been designing and installing isolating floating floors in gymnasia for decades and there is a catalogue of data illustrating reductions in noise which can be achieved. However, applying prior data to new projects without knowledge of the underlying structure cannot be done, plus site testing has further uncertainty due to size of the test surface and constrained air effects.

5. Tests on a representative floating floor were commissioned to better understand the underlying dynamics, with a view to producing a validated dynamic model. This paper describes the testing carried out and results which provide a route to a prediction model, the ultimate aim of which is to provide certainty on the isolation provided for future projects through the use of a scalable and validated model.

 

2. BACKGROUND TO REQUIREMENT

 

1. The principle of isolating noise and vibration from gym activities is one of physics. The isolation system used needs to attenuate frequencies of energy which would otherwise excite or be carried by the structural slab into other structural elements. Ultimately, any employed attenuation measures should be sufficient to avoid risk of disturbance elsewhere in the building.

2. The principles of impact isolating floors have been established for decades. Structural floors can be made stiffer and thicker to make them less responsive to impact, but this adds mass and cost which is inefficient relative to the acoustic benefit. Introducing a separating air gap tremendously increases isolation, far more than can be achieved with a much thicker structural slab saving space, materials, and cost. Supporting a floating floor with correctly compressed and selected isolators results is key to a well performing system explained by the physics of transmissibility.

3. The criticality of the air gap is why most clients insist on a jack-up system which involves constructing the floor directly on the structural floor and ‘jacking’ once cured. This provides certainty of the correct air gap, removal of formwork layers susceptible to humidity and the ability to adjust the compression of every isolator irrespective of floor level.

4. Test data are well established for projects focussed on the control of audible sources, such as cinema, studios, and entertainment venues. Jack-up floating floors using natural rubber isolators are highly effective as natural frequency of support can be achieved below 10Hz, transmissibility theory shows that isolation will begin at √2 above the natural frequency so well below the threshold of hearing. Dynamic stiffness of the isolators is key, needing to be part of the calculation of natural frequency.

5. All elastomeric isolators, including natural rubber isolators using a low dynamic stiffness formula have damping which introduces slight lag on response. This is a functional limitation under impact since the duration of a pulse is small, meaning common for energy at lower frequencies to be amplified while higher are damped. Since this commonly marries with structural response frequencies switching to undamped springs reduces this effect.

6. However, the stiffness of the supporting structure still dominates. Physical testing removes some risk but a method of predicting performance is not currently viable. For this reason, Hoare Lea were approached to carry out characterisation testing of a floating floor with an ultimate aim of producing a scalable and validated prediction method.

 

3. DESIGNING THE RESEARCH

 

1. Testing was carried out at in the basement of Hoare Lea offices, Manchester. The venue was chosen as the ground bearing basement slab, removes dynamic influences which would exist with suspended slabs. This allows the data to clearly establish the characteristics of the floating floor in order to generate clean insertion loss data.
2. It was important to understand the effects of different isolators, so the floor was designed with this in mind. The floating floor is a typical concrete jack-up using the Mason FS system. The floor is a 3.2m x 1.6m reinforced concrete (RC) slab with a constant thickness of 0.1m (100mm). The floor is suspended via 6 no. springs, which can be removed and replaced via jacking the floor and switching the spring component accordingly. The isolator locations can be identified in Figure 1 as numbers enclosed within circles ranging from 1-6. Circles enclosing the number 7 refer to damping elements which can be engaged/disengaged as needed. Information of the isolator types used in tests summarised in this document are provided Figure 2.

 

 

Figure 1: Layout of test floor

 

 

Figure 2: Isolation selection and section

 

3. For reasons of focussing on the performance of specifically the floating floor, the tests were designed around insertion loss rather than transmission loss, which required baseline tests followed by different gym activities carried out on the two spring configurations of interest, 25mm and 50mm compression springs.

   1. Baseline – on the base slab (on grade) with no floated slab on top

   2. 50mm spring – vibration measurements on base slab with floated slab on 50mm springs installed

   3. 25mm spring – vibration measurements on base slab with floated slab on 25mm springs installed

4. Measurements of vibration levels were conducted with calibrated multichannel analysers and accelerometers. The location of the accelerometers was carefully chosen next to the coupling locations of the springs and dampers with the base slab. These positions remained unchanged for each test condition as Insertion Loss measurement requires levels at the receiver (base slab). The test schematic with sensor locations is shown in Figure 3.

 

 

Figure 3: Accelerometer locations

 

5. A combination of 100 mV/g and 500 mV/g sensitivity accelerometers was used to capture the response of the base slab as a result of generated impacts on floated slab. Levels within each test were captured up to 10 kHz one-third octave bands. In practise, structure borne sound frequencies of interest are confined to low and mid frequency regions and as such, results reported herein are limited to a maximum of 5 kHz.
6. For each test, multiple iterations (e.g. up to 4 impacts per dumbbell type and up to 3 repeated runs of bike/treadmill) were conducted. For impact sources (basketball, dumbbell), the maximum vibration level within each impact was extracted. These were dropped using a drop tower rig to enhance the repeatability of the tests. The maximum impact levels within each hits/run were then averaged to provide the insertion loss as shown in Equation 1.

 

 

7. Where, L_a,b is the spatially averaged vibration acceleration level during baseline tests, and, L_a,s is the spatially averaged vibration acceleration level during tests with floated system on top (with 50mm or 25mm deflection springs).

8. In the case of treadmill and exercise bike tests, which are steady-state sources, the vibration levels were instead averaged over the run-duration. Corresponding IL results were calculated as per the equation above. All IL results up to 5 kHz bands calculated also consider the limiting effects of existing background levels, so that a usable frequency range could be established, as explained next.

9. Frequency range of assessment. Any results obtained from tested sources are subject to background levels to ensure the results obtained in each frequency band are reliable and not adversely limited by background. As such two checks were made,

   • Background levels on floated slab – Levels on the floated slab under each impact/run were compared against the background levels on the floated slab. Sample accelerometers were installed on the floated slab for each test to allow this.

   • Background levels on base slab – Levels on base slab under each impact/run were compared against the existing background levels on the base slab. Accelerometer positions shown within Figure 5 were used for this purpose.

10. In each of the comparisons above, a Signal-to-Noise Ratio (SNR) ≥ 6dB was used as a criterion to separate the frequencies of interest from the background noise regions. Based on this, the frequency range of assessment was established for different sources (dumbbell, basketball and bike/treadmill sources, Table 1). This process was carried out for each individual test type to assess the relevant frequency range as defined by SNR>6dB.

11. Furthermore, care was taken during testing and post-processing to discount any extraneous vibration effects from adjacencies. Any vibration from the test setup elements was minimised as far as practically possible using elastomeric pads below frame feet, smooth ropes, etc.

 

 

Table 1: Test schedule and content

 

4. FINITE MODELLING AND VALIDATION

 

1. Frequency Response Function (FRF) measurements were taken to validate the model. Excited by force hammer, the response was measured in several locations. Tests were carried out using 50mm springs, damping elements disengaged and the force hammer was applied centrally to the floor. Results are shown in Figure 4. The principle natural frequency of the system can be seen and also further peaks representing other modal responses.

2. The finite element model was constructed on the basis of a simple representation of the slab and omits the structural steel reinforcement. Springs are modelled as idealised spring elements, as opposed to direct modelling of the spring geometry. Only the vertical component of spring stiffness was considered.

 

 

Figure 4: Initial FEM results showing modal shapes

 

5. RESULTS – FINITE ELEMENT VALIDATION

 

1. The results of the modal analysis have been used to determine the FRF of the slab in an additional assessment of harmonic response. In this assessment, a notional 1N harmonic force was used to excite the model in an area corresponding to the hammer strike location used in physical measurements. Response locations were chosen to match those also selected for physical testing. A typical response is shown below in Figure 5.

 

 

Figure 5: Measured data plotted against FE simulation

 

2. It can be seen that there is good agreement between measured and predicted FRF response. The significance of this finding is that the slab dynamics can be sufficiently modelled in a prediction exercise. With added modelling of the receiver system, a ‘Virtual Acoustic Prototype’ could be built to predict the performance of a system beforehand to optimise product selection on a case-by-case basis.

 

6. RESULTS – INSERTION LOSS TESTING

 

1. A subset of results is presented in this section for the sake of brevity. Figure 6 shows results recorded for 25mm springs under a variety of weight drops, different heights and masses. Consistent energy losses can be seen across the spectrum which is reflective of site testing.

 

 

Figure 6: Measured insertion loss data of weight drops

 

7. FURTHER WORK

 

1. The result of the initial work is highly encouraging. The correlation between FEM and measured data indicates that with further validation work on different sized floors and on suspended slabs will permit a model to provide accurate results. The next steps are to carry out baseline testing and further testing along lines presented in this paper on several real-world installations. The FEM will be refined with the ultimate aim of a scalable model which when fed detail on supporting structure can provide reliable insertion loss values to the consulting acoustician.

 

8. CONCLUSIONS

 

1. It is concluded that this work is a success and that is a good basis for further work that should result in a useful tool to industry. We have a validated FEM which has shown to be accurate when subjected to a number of typical gym activities.

2. The complete work includes data on damping elements and the effects of covering layers but there is insufficient space to present this here. It is critical that these factors be considered considering the wide range of finishes and business models which exist in the health and fitness industry.

 

9. ACKNOWLEDGEMENTS

 

1. We wish to thank Martin McNulty and Nikhilesh Patel of Hoare Lea Manchester for their continued support in designing and carrying out these tests. We look forward to developing the results into a useable tool in the near future.

 

---

¹ adam@masonuk.co.uk