Measuring the impact force from the ISO impact ball and comparison with the tapping machine and alternate input method Sunit Girdhar 1 Jason Blough Michigan Technological University Houghton, MI Andrew Barnard Pennsylvania State University John LoVerde Wayland Dong Veneklasen Associates, Inc.

ABSTRACT The impact ball was recently standardized within the ISO standards as a low-frequency input impact source for impact testing of floor-ceiling assemblies. The input force due to an impact ball is not measured during the tests. A force measurement plate was created with three force transducers to measure the impact force due to ball drops on six different floor assemblies. The input force showed a really good comparison for heavy and lightweight floors in lower frequency but poor force excitation for all the floors above 100 Hz. The input force values were compared with a modified tapping machine’s force input levels showing the tapping machine has poor low-frequency excitation (in comparison) but an improved high-frequency excitation. An alternate input method was developed where a force transducer is added to the impactor to make live measurements of the input force and the impact tip can be changed to modify the range of targeted frequencies for any test.

1. INTRODUCTION

With humans spending more than 90% of their time indoors [1], it is imperative to create a comfortable indoor space, primarily, the places of residence. With the increased development of multistory buildings across urban cities, it is hard to find quiet spaces for people, sometimes even in their own homes. Footstep noise is considered the most annoying noise source in multistory residential buildings [2-4]. A standard tapping machine (first adopted in Germany in 1938 [5, 6]) is used within the ISO standards [7-9] to evaluate the footstep impact noise performance of different floor ceiling assemblies.

According to the standard, the tapping machine is placed at four random locations on the floor in the source room such that no location should be closer than 0.5m from the room boundaries. The radiated Sound Pressure Levels (SPL) are measured using fixed microphones or a scanned microphone, with restrictions on the microphone locations as compared to the room boundaries and

1 sgirdhar@mtu.edu

themselves. The energy averaged SPL are compared to a reference curve to calculate the single number rating of the floor. The measurements are made in One-Third Octave (OTO) frequency bands ranging from 100 – 3150 Hz with an option to go as low as 50 Hz and as high as 5000 Hz in frequency bands, if desired.

Research shows that the tapping machine impacts do not correlate well with footstep impacts [4, 10-12] and the results from the tapping machine tests fail to correlate with the subjective response of the residents. Mariner et al. [6] showed that two assemblies with similar single number ratings using a tapping machine had a difference of 11.5 points in their subjective ratings. Olynyk et al. [13] showed that two floors with a subjective rating difference of twenty points had only a two point difference in their impact rating. In another work, two assemblies with an 8-dB spread in their impact rating had the same subjective response [14].

This led to the development of an impact ball as a low-frequency impact source to mimic heavy- soft impacts. The ball was reverse engineered based on the peak input force levels and the duration of the impact (that defines the frequency range excited). This ball had a higher low-frequency input levels as compared to the tapping machine but a lesser frequency excitation range [15]. In some early studies, the impact ball shows a much better correlation with the children jumping and running as compared to the tapping machine [15, 16]. This impact ball is now a part of the ISO standards to be vertically dropped in a free fall from a height of 1 m for at least four locations on the floor under test. This test can be done from 50 – 630 Hz OTO bands according to the ISO standard [9].

The standard requires an impact ball to generate a specified maximum force exposure level but it is unclear how this is measured. During an input test, this input force is not required to be measured, but the sound radiated from the assembly depends on the input force, floor compliance, and the vibration propagation from the floor to the ceiling (and flanking paths, which are out of scope of this work). Therefore, the input force during a standard test is important understand the performance of floor-ceiling assemblies due to impacts. Olsson et al [17] created a force measurement rig for impact ball tests using a single force transducer and showed that the force spectra is similar regardless of the type of floors below 55 Hz but starts to show deviations above that frequency based on the type of the floor. Our work for this project is inspired by this previous work to further understand and evaluate the input force characteristics of the impact ball and the dependence of this force on the floor compliance.

A force plate was created to directly measure the input force due to impact ball drops. These input forces were compared with the force levels from a modified tapping machine based on previous work [18]. This was done to better understand how the floor behaves due to the two different types of standard impact methods. Based on the type of construction of these two impact methods, we expect to see higher force levels with the impact ball in low-frequencies but a wider frequency excitation range with the tapping machine.

In addition to the standard impact methods, we will also introduce an alternative input method developed by an in-house student design team. The impactor developed by the team is equipped with a force transducer so live force level measurements can be made. In addition to this, the impactor is highly customizable to get the desired input force levels in a certain floor. More details about this alternate input method would be shared in a later section. 2. METHODOLOGY

The force plate was designed and manufactured in-house and the input force using the force plate was calculated using Fast Fourier Transform (FFT) method. This section describes the force plate, signal acquisition parameters, signal processing procedure, the floors tested with the impact ball, and gives a summary of the work done to measure input force from the tapping machine (based on the previous work).

2.1. Force plate design and manufacturing The force plate is made using Aluminum 6061 with a thickness of 12.7 mm (0.5 in) and a diameter of 127 mm (5 in), shown in Figure 1 (a). In comparison, the ball diameter is approximately 180 mm

(7 in), shown in Figure 1 (b). The force plate is equipped with three force transducers (PCB 208A03 SN4830, SN4831, and SN5316), shown in Figure 1 (c). Figure 1 (a, b, and d) shows the force plate in assembled conditions on the floors by hot gluing the force transducers to the floors.

Figure 1 (a) The force plate as assembled on one of the floors (b) Force plate compared to the standard impact ball for size comparison (c) The three transducers on the force plate (d) The force plate as assembled on one of the floors

The force plate was simulated in Simcenter 3D and an experimental modal analysis was performed to verify that the dynamic modes of the force plate would not affect the frequency range of interest (below 500 – 1000 Hz).

2.2. Data acquisition parameters Simcenter Test Lab Impact Testing module was used for data acquisition with a frequency resolution of 0.5 Hz and the bandwidth of measurement was 4096 Hz. Uniform windows were used on all force channels. Seven averages were recorded. A trigger was set on one of the force transducer channels for 50 N with a 0.1 second pre-trigger to measure the entire impact in one acquisition period of 2 seconds. The time domain throughput data was also recorded during the test.

2.3. Data processing procedure Data processing was performed in MATLAB using the exported time-domain throughput data from Simcenter Test Lab. Using the same trigger settings as Simcenter Test Lab, the raw time domain data were divided into 2 seconds worth of data on all the channels. The total input force in time domain is simply the sum of all three force channels. An FFT algorithm was performed on the individual force channels and the overall force spectra was calculated as the addition of all the force channels. The linear autopower was the magnitude of the overall force spectra.

2.4. Details of the floors tested Five floors were tested with the impact ball for this work. These floors are shown in Figure 2 and the details are given below:

• Two heavyweight reinforced 6-in (152 mm) concrete floors (HC1 and HC2)

• Two lightweight joist-framed floors with hardwood finish (LH1 and LH2) • One lightweight joist-framed floor with carpet tile finish (LC1)

Figure 2 For this work, two heavyweight reinforced 6-in concrete floors (HC1 and HC2), two lightweight joist-framed floors with hardwood finish (LH1 and LH2), and one lightweight joist- framed floor with caret tile finish (LC1) were tested.

2.5. Tapping machine force measurement (based on previous work) The standard tapping machine as defined in the ISO standards does not have a provision to measure the input force during a floor impact test. In a previous work [18], the authors modified a standard tapping machine by adding a force transducer in the middle of the hammer head (standard) and the hammer shaft (modified), as shown in Figure 3 (a) and (b). Only a single hammer (out of five) was used for simplicity, as shown in Figure 3 (c). A single impact (instead of continuous impacts) was performed with the tapping machine and the data acquisition parameters were the same as those for the impact ball test mentioned earlier.

Figure 3 (a) The tapping machine hammer with a force transducer to measure input force (b) a close- up view of the force transducer showing the location between the tapping machine hammer head and the hammer shaft, and (c) the tapping machine on one of the floors tested. Four out of the five hammers were removed for simplicity.

3. RESULTS

In this section, we will discuss the impact ball force results in the time domain, frequency domain, and compare these results with the tapping machine.

3.1. Time domain results from the impact ball test

In the time domain, the total input force due to the impact ball drop is simply the sum of all three force transducers. This input force values for all the five floors tested with seven impacts each are shown in Figure 4. For all the five floors tested, the peak input force levels range from 1600 – 1850 N (1.26 dB), regardless of the type of floor construction. The approximate time the impact ball stays in contact with the floor is 15 msec. The x-axis is normalized to avoid any confusion with the time stamp on the data acquisition system. Comparing the measured force levels to the ideal impact ball spectra according to Annex A in ISO 16283-2 [9], the peak impact ball force level is approximately 1450 N with an impact duration of 20 msec.

Figure 4 The peak input force due to the standard impact ball drops show that the peak force ranges from approximately 1600 – 1850 N (1.26 dB variation) regardless of the type of construction of the floor. The normalized time axis (x-axis) shows that the approximate time of the impact is 15 msec.

The input force due to the standard impact ball is compared with the tapping machine force in Figure 5 for all the five floors. The tapping machine peak force values are on the left side of the plot and the impact ball peak force values are on the right side with a common y-axis. The x-axis on the left and the right side are both normalized for easy comparison. For the tapping machine, the peak force level ranges from approximately 300 N to 5000 N (25 dB difference) based on the type of floor construction, unlike the impact ball. The peak levels for the tapping machine for the HC floors is approximately 9 dB higher than the impact ball, but the peak levels are 8 – 15 dB lower for the lightweight floors (LH and LC floors). The width of the impact with the tapping machine is much smaller than that of the impact ball. Based on this width, we expect to see the tapping machine impacts to have a wider frequency excitation range as compared to the impact ball. As a reminder, the tapping machine force data comes from previous work [18].

Force input levels with the impact ball 2000 —H¢1 1800 1600 1400 1200 Force amplitude (N) +200 -0.03 0.02 -0.01 0 0.01 0.02 0.03 Normalized time (s)

Figure 5 The time domain force comparison between the tapping machine (left) and the impact ball (right) shows that the peak force values for the tapping machine varies by approximately 25 dB based on the floor construction for the tapping machine, unlike the impact ball. The peak force levels are approximately 8 – 15 dB lower for the LH and LC floors with the tapping machine as compared to the impact ball.

3.2. Freq domain results from the impact ball test The overall force in the frequency domain for all the floors tested with the impact ball are shown in Figure 6 with force in dB on the y-axis and the frequency in log scale on the x-axis. Below 100 Hz, the force spectra of all the floors are within 1 dB, regardless of the type of floor construction. This seems to be in-line with the time domain impact levels shown in Figure 4. At 100 Hz, the force spectra for all the floors falls by approximately 35 dB. This means that above 100 Hz, the Signal to Noise Ratio (SNR) of the impact force on the floors may be poor. Recall that the ISO standard [9] allows the impact ball to be used at frequencies below 630 Hz OTO band. This is represented by the black colored vertical dashed line with on the plot. The frequency spectra falls by more than 50 dB (approximately) at 630 Hz meaning that the SNR might be poor in parts of the measurement frequency range allowed by the standard.

Force input levels for the tapping machine and the impact ball 500! Tapping machine Impact Ball oor 0 oor 0 (Ot Normalized time (s)

Figure 6 The frequency domain force spectra for all the floors tested is within 1 dB below 100 Hz and it falls by approximately 35 dB at 100 Hz. This signals that the SNR may be poor for the impact ball above this frequency. At 630 Hz OTO band (allowed by the ISO standard for the impact ball), the spectra falls by more than 50 dB.

Figure 7 compares the frequency domain force levels for the tapping machine (solid lines) with the impact ball (dashed lines) below 707 Hz (the upper limit for the 630 Hz band, the highest allowed within the ISO standard for the impact ball). The frequency is plotted with a log scale in the x-axis and the force is shown in dB (ref 1N). The low-frequency force levels with the tapping machine varies by approximately 2 – 3 dB based on the type of floor construction and it is approximately 25 – 30 dB lower than the impact ball force levels. The force spectra for the tapping machine falls off much slowed than the impact ball (similar to the conclusions drawn from Figure 5). At 707 Hz (upper limit of the 630 Hz OTO band), the impact ball force levels are approximately 18 – 19 dB lower than the tapping machine for the HC floors.

Input force (dB) with the standard ISO impact ball for all the floors tested & é Input Force Spectra (dB ref 1 N) ® 8 -50 ~60 -80 10 107 10° Freq (Hz) - log scale

Figure 7 Comparison between force spectra with the tapping machine (solid) and the impact ball (dashed) shows that the low-frequency force levels for the tapping machine are approximately 25 – 30 dB lower than the impact ball but the tapping machine has a wider frequency excitation bandwidth as compared to the impact ball, especially for HC floors.

4. ALTERNATE INPUT METHOD DEVELOPMENT

An alternate input impactor was developed by a senior design team at Michigan Technological University in collaboration with the authors to address some of the disadvantages of the existing standard input methods. This prototype machine is shown in Figure 8 (left), and it uses a slider crank mechanism (Figure 8 (right)) to drop a mass from a certain height. The impactor is equipped with a force transducer and an impact tip. Both the impact tip, and the initial height of the mass can be changed to control the force injected in any given floor assembly. The slider crank mechanism is controlled by a motor and a clutch that helps achieve the desired height of the mass, and prevents any rebound by catching the mass just after the impact. The impactor generates a single impact (unlike the continuous impacts with the tapping machine) to generate a smooth frequency domain spectra. The power supply runs using a laptop style power cord, making it really easy to work globally with a simple power cable. The machine has its own password protected Wi-Fi network

Input force (dB ref 1 N) gqftbut force spectra with the tapping machine(solid) and impact ball(dashed) © PS HS FS SLHKS S £ CSS Freq (Hz) - log scale

that lets the operator connect to the on-board computer using their mobile phones that can be used to set the impact height and to trigger the machine to enable remote operation.

Figure 8 The alternate impact machine developed at Michigan Technological University (left) uses a crank slider mechanism (right) to drop a mass from a certain height. The impact tip and the drop height can be changed to control the force injected into the surface during an impact test. The impactor is equipped with a force transducer for live measurements of the input force

More details from this alternate impactor and the data collected for some of the floors was recently presented at the Noisecon 2022 meeting in July 2022. Please refer to the conference proceedings from this meeting for more details. 5. CONCLUSIONS AND FUTURE SCOPE

With all the floor constructions tested, two heavyweight and three lightweight (with hardwood and carpet finish), the time domain force levels with the impact ball vary within 1 – 2 dB and the low- frequency force levels vary by less than 1 dB. This is unlike the results we saw previously from the tapping machine [18]. The low-frequency levels with the impact ball are approximately 25 – 30 dB higher as compared to the tapping machine, as expected based on the construction of the rubber impact ball and steel tapping machine. However, the tapping machine has a wider frequency excitation range as compared to the impact ball, as also seen with the width of the impact of the two standard input methods. For the impact ball, the force spectra falls by approximately 35 dB at 100 Hz and approximately 50 dB at the 630 Hz OTO band (highest measurement band allowed according to the ISO standard). This signals that the SNR for the impact ball might be poor in some of the frequency bands defined in the ISO standard. This may need further evaluation.

In the future, we want to measure the floor impedance with the impact ball for some of these floors and compare them with the floor impedance measured with the tapping machine during earlier work. The impedance measurement will also give us an indication of the coherence during the measurement, which will further help us understand the SNR of the impact ball drop above 100 Hz. We are also looking into constructing a Printed Circuit Board (PCB) with a summing op-amp to directly add the analog signal from the three force transducers so all the calculations can be directly done in the Simcenter Test Lab software instead of doing this manually in MATLAB using the time domain data. We are also exploring different ideas to measure the sound radiation in the receiving room during a standard impact test with improved reproducibility, especially at low-frequencies where the sound field is generally non-diffuse. To do this, we are first creating a simulation model to try out some ideas and this will be followed by real-world testing.

6. ACKNOWLEDGEMENTS

The authors gratefully acknowledge the Paul S Veneklasen Research foundation for their support with this work and the travel to this conference. We would also like to thank our colleagues and faculties at the Dynamic System Laboratories at Michigan Technological University, and our colleagues at Pennsylvania State University for their support and all the insightful conversations during this research. The authors also want to thank the Michigan Tech Senior Design Team that tirelessly worked on the alternate impact method and the team mentors for guiding the students throughout the project. The authors also want to thank the organizers of the InterNoise 2022 conference for their efforts. 7. REFERENCES

1. Emily Anthes The Great Indoors , Farrar, Straus, and Giroux, 2020. 2. Kaj Bodlund, Alternative reference curves for evaluation of the impact sound insulation between

dwellings, Journal of sound and vibration , 102(3) , 381-402, (1985). 3. E Gerretsen, A new system for rating impact sound insulation, Applied Acoustics , 9(4) , 247-263,

(1976). 4. B. G. Watters, Impact ‐ Noise Characteristics of Female Hard ‐ Heeled Foot Traffic, The Journal of

the Acoustical Society of America , 37(4) , 619-630, (1965). 5. H Reiher, Über den Schallschutz durch Baukonstruktionsteile (Sound Isolation by Building

Construction Elements), Beih. Ges.-Ing , 2(11) , 2-28, (1932). 6. T Mariner and HWW Hehmann, Impact ‐ Noise Rating of Various Floors, The Journal of the

Acoustical Society of America , 41(1) , 206-214, (1967). 7. 717-2: Acoustics -- Rating of sound insulation in buildings and of building elements -- Part 2:

Impact sound insulation , ISO, (2013). 8. 10140-3: Acoustics—Laboratory measurement of sound insulation of building elements. Part 3:

Measurement of impact sound insulation , ISO, (2021). 9. 16283-2: Acoustics — Field measurement of sound insulation in buildings and of building

elements — Part 2: Impact sound insulation , ISO, (2018). 10. A. C. C. Warnock, Low frequency impact sound transmission through floor systems, INTER-

NOISE and NOISE-CON Congress and Conference Proceedings , 1992(2) , 743-748, (1992). 11. Wanqing Shi, et al., A comparison of the characteristics of different impact force with a standard

tapping machine, INTER-NOISE and NOISE-CON Congress and Conference Proceedings , 739- 742, (1995). 12. N. Amiryarahmadi, et al., Identification of Low-Frequency Forces Induced by Footsteps on

Lightweight Floors, Acta Acustica united with Acustica , 102(1) , 45-57, (2016). 13. D. Olynyk and T. D. Northwood, Assessment of Footstep Noise through Wood ‐ Joist and

Concrete Floors, The Journal of the Acoustical Society of America , 43(4) , 730-733, (1968). 14. D Olynyk and TD Northwood, Subjective Judgments of Footstep ‐ Noise Transmission through

Floors, The Journal of the Acoustical Society of America , 38(6) , 1035-1039, (1965). 15. Hideki Tachibana, et al., Development of new heavy and soft impact source for the assessment

of floor impact sound insulation of buildings, INTER-NOISE and NOISE-CON Congress and Conference Proceedings , 1998(4) , 1227-1230, (1998). 16. Jin Yong Jeon, et al., Use of the standard rubber ball as an impact source with heavyweight

concrete floors, The Journal of the Acoustical Society of America , 126(1) , 167-178, (2009). 17. Jörgen Olsson and Andreas Linderholt, Measurements of low frequency impact sound frequency

response functions and vibrational properties of light weight timber floors utilizing the ISO rubber ball, Applied Acoustics , 166 , 107313, (2020). 18. Sunit Girdhar, et al., Measuring the force due to standard tapping machine and floor impedance

for ASTM standards, INTER-NOISE and NOISE-CON Congress and Conference Proceedings , 263(6) , 767-777, (2021).