• For Contributors +
• Journal Search +
Journal Search Engine
ISSN : 1598-4095(Print)
ISSN : 2287-7401(Online)
Journal of The korean Association For Spatial Structures Vol.19 No.1 pp.117-126
DOI : https://doi.org/10.9712/KASS.2019.19.1.117

# Comparison Between Performance of a Sound-Triggered Measurement and an Amplitude-Triggered Measurement in Shaking Table Tests

S. T. Mapungwana*, Jong-Ho Lee**, Sung-Won Yoon***
*Dept. of Architecture, Seoul National University of Science and Technology
**Dept. of Architecture, Seoul National University of Science and Technology
교신저자, 정회원, 서울과학기술대학교 건축학부 교수, 공학박사 School of Architecture, Seoul National University of Science and Technology Tel: 02-970-6587 Fax: 02-979-6563 E-mail: swyoon@seoultech.ac.kr
January 15, 2019 February 13, 2019 February 13, 2019

## Abstract

Micro-Electro-Mechanical Systems (MEMS) sensors have been widely used in Structural Health Monitoring due to their convenience and lower costs in comparison to conventional sensors. Triggered measurements are relevant in events such as earthquakes because unlike continuous measurements, they only record the structural response once an event happens. This is more cost effective and it makes the data more manageable because only the required measurements from the event are recorded. The most common method of triggering is amplitude triggering. However, lower input amplitudes (less than 0.1g) cannot be triggered by using this method. In this paper, sound triggering was introduced to allow triggered measurements for lower input amplitude values. The performance of the sound triggering and amplitude triggering were compared by a series of shaking- table tests. It was seen that sound- triggering method has a wider frequency (0.5~10Hz) and amplitude (0.01~1.0g) range of measurements. In addition, the sound triggering method performs better than the amplitude triggering method at lower amplitudes. The performance of the amplitude triggering, in terms of the triggering being simultaneous improves at higher input amplitudes.

# 진동대를 이용한 모바일 진동 계측 기기의 사운드 트리거 계측과 진폭 트리거 계측 성능 비교

마 푼과나 시부시시웨*, 이 종 호**, 윤 성 원***
*학생회원, 서울과학기술대학교 건축과, 석사과정
**학생회원, 서울과학기술대학교 건축과, 박사과정

## 초록

National Research Foundation of Korea
2016R1A2B2014064

## 1. Introduction

Micro-Electro-Mechanical Systems (MEMS) sensors have been widely used in Structural Health Monitoring (SHM) based on structural vibration measurements due to their smaller and convenient sizes and their considerably lower prices than conventional sensors. Smart devices have become a possible option for acceleration measurements, data storage and cloud sharing1). This is due to the advancement in the central processing unit of in-built MEMS sensors in the devices.

Small-scale shake table tests have been carried out to investigate the smart phones’ capability of measuring vibrations at different amplitudes and frequencies2). Comparison between the performance of smart phone devices with in-built MEMS sensor and a reference sensor in shaking-table tests has been done3). It was seen that the performance of MEMS sensors was generally good in comparison to an ICP sensor.

In seismic event measurements, it is more relevant to have triggered measurements because unlike continuous measurements, a triggered measurement only measures the structural response once an event occurs. This is cost-effective and makes it easier in data management. The commonly used method of triggering is amplitude triggering. Amplitudetriggered measurements were conducted by using smart phone devices connected at different levels. However, for small intensity ground-motion input, the devices in lower-story levels could not be triggered because the vibration amplitude was smaller than the threshold amplitude1).

The minimum acceleration level that could trigger the smart phone devices was 0.1g in any of the three axes4). This means that acceleration levels below 0.1g cannot be measured by amplitude-triggering.

However, by using sound triggering, measurements for acceleration levels lower than 0.1g can be done. In this study, the soundtriggered measurements for both acceleration levels below 0.1g and above 0.1g was introduced.

The measurements of acceleration levels above 0.1g are significant in comparing the performance of the sound-triggered measurements and amplitude-triggered measurements. This study hopes to investigate the accuracy of using sound-triggered measurements by using a vibration measuring application called Vibration Application for measuring vibration response in small-scale shaking table tests.

## 2. Vibration application

Vibration Application is an application for iPhones, iPods and iPads which analyzes the vibratory and rotational motions by using an accelerometer and gyroscope found inside the devices5). The MEMS accelerometer analyzes the vibrational motion and the gyroscope analyzes the rotational motion. The motions can be recorded in each of the three axes as shown in <Fig. 1>. An important feature of the Vibration Application is its ability to take measurements from triggered events. The Vibration Application can be triggered by amplitudes or by sound.

The 3-axis MEMS accelerometer (BMA280, Bosch Sensortec) can measure within a range of ± 2g7). In addition, the Vibration Application can measure the sampling rate in the range of 0~100Hz.

### 2.1. Amplitude trigger

The minimum amplitude that can trigger the measurements is 0.1g. Any of the channels can be measured from triggered events. To set a trigger, the desired channel should be turned on and the slider can be adjusted to the trigger value. For this experiment, the trigger value was set to 0.1g and only the x-axis channel was turned on.

### 2.2. Sound trigger

Due to the advancement of the Vibration Application, the internal microphone was added as an input source in April 20155). This enabled the Vibration Application to take measures triggered by sound. To enable the sound triggering, the microphone trigger should be turned on and a desired value of minimum sound level can be set by moving the slider. The range of minimum sound level that can trigger the device is from -50.0dB to 0.00dB. In this experiment, the minimum sound trigger value was set to -8.83dB.

## 3. Small-scale shaking table tests

To investigate the devices’ performances in trigger measurements for different frequency and amplitude levels, a uniaxial shaking table was used. <Fig. 2> shows the equipment used, and the specifications of the shaker is given in <Table 1>. Details of the Power Amplifier used are also given in <Table 1>.

### 3.1. Specimen used

Four apple devices with in-built MEMS sensors were used for the experiment. <Table 2> shows the specifications of the four apple devices used. The sufficient sampling frequency for MEMS sensor is 100Hz.

### 3.2. Experimental setup

<Fig. 3> shows the layout of the specimen used on the shaking table. The APS 400 uniaxial shaking table was used to produce sin wave excitation for the experiment and the APS 145 amplified the input excitation.

The Keysight 33210A waveform generator served sinusoidal waves to the shaking table. A total of 18 tests were carried out for different input amplitudes and frequencies as shown in <Table 3>.

The input frequency ranged from 0.5Hz to 10Hz and the input amplitude ranged from 0.01g to 10g. Since amplitude trigger measurements cannot be done for amplitudes less than 0.1g, tests for amplitudes less than 0.1g were only for the sound-triggered measurements.

## 4. Shaking-table test results

The response of the shaking table was analyzed by plotting time history curves. The natural frequency of the system was obtained by carrying out Fast-Fourier Transformation (FFT) to obtain the power spectrum.

### 4.1. 0.5Hz input frequency

Two tests (input amplitudes of 0.01g and 0.05g) were carried out at input frequency of 0.5Hz.

Time response of the sound-triggered measurements was obtained as shown in <Fig. 4>. The time history response shows results for the first 20 seconds. From <Fig. 4>, iPhone 1 tends to delay in acceleration measurements. This observed time lag might be due to deviation in sampling frequency during measurements. It can also be seen from <Fig. 4 (b)> that the iPad 2 and iPhone 2 overestimate the amplitude measurements.

<Fig. 5> shows the power spectra for sound-triggered measurements for the input amplitudes 0.01g and 0.05g at 0.5Hz. The power spectrum was analyzed using all of the data for a time duration of 40 seconds.

### 4.2. 1.0Hz input frequency

Three tests (input amplitudes of 0.01g, 0.05g and 0.1g) were carried out at input frequency of 1.0Hz. Time response of the sound-triggered measurements was obtained as shown in <Fig. 6>. The time history response shows results for the first 10 seconds for a better visual of the graphs. Compared to the 0.5Hz frequency measurements, the time lag has reduced but it can still be observed from <Fig. 6 (b)> that iPad 2 and iPhone 2 tend to overestimate the acceleration values. Fig. 7

By comparing <Fig. 8> and <Fig. 9>, a slight time delay of iPhone 2 can be observed for amplitude-triggered measurements. This might be because at low input amplitudes (0.1g) and frequencies, the devices cannot be easily amplitude-triggered hence there might have been a delay in triggering of the iPhone 2 at 0.1g.

<Fig. 10> and <Fig. 11> show the corresponding power spectra for sound-triggered and amplitude-triggered measurements respectively. The power spectrum was analyzed using all of the data for a time duration of 40 seconds.

### 4.3. 5.0Hz input frequency

Three tests (input amplitudes of 0.1g, 0.5g and 1.0g.) were carried out at input frequency of 5.0Hz. Time response of the sound-triggered measurements and amplitude-triggered measurements were obtained as shown in <Fig. 12> and <Fig. 13> respectively.

For a better visual of the graphs, the x-axis was cut to the first 2.5 seconds. As input amplitudes increase, the time lag in amplitude-triggered measurements reduces as seen from <Fig. 13 (b)> and <Fig. 13 (c)>. This shows that amplitude triggering performs better at higher input amplitudes in comparison to sound triggering.

<Fig. 14> and <Fig. 15> show the corresponding power spectra for sound-triggered and amplitude-triggered measurements respectively. The power spectrum was analyzed using all of the data for a time duration of 40 seconds.

### 4.4. 10.0Hz input frequency

Three tests (input amplitudes of 0.1g, 0.5g and 1.0g) were carried out at input frequency of 10.0Hz. As input amplitudes increase, the time lag also reduces. Measurements show similar results with 5.0Hz input amplitude, so plotted graphs were omitted.

## 5. Analysis of shaking-table test results

### 5.1. Analysis of error rate in amplitude data

To analyze the accuracy of the experiment, the error rate (ψa) was calculated using equation (1) where Apeak is the average peak amplitude and Ainput is the input amplitude3).

$ψ a = A p e a k − A i n p u t A i n p u t × 100$
(1)

The range of error values for amplitudetriggered measurements is 0.03~7.83% and the range of error values for sound-triggered measurements is 0.03~17.5%. The high error values (3.90~17.5%) for sound-triggered measurements is mostly from low input amplitude measurements (0.01g and 0.05g). For input amplitudes greater than 0.1g, the range of error values is 0.03~6.73% with most of the error values below 3% error except for anomalous results at 0.1g for 5.0Hz and 10.0Hz. For both amplitude-triggered and sound-triggered measurements, these anomalous results are observed for measurements from iPhone 2.

The high negative error values (-7.83~-6.14%) show an overestimation of the measured acceleration response. These high error values might be due to the specifications of the in-built MEMS sensor of the iPhone 2. From <Table 2>, it can be seen that iPhone 2 has a lower sensitivity (16LSB/g) compared to other MEMS sensors (4,096LSB/g).

A general trend of decrease in error as input amplitude increases is also observed. This is because MEMS sensors are more accurate in vibration measurements at higher input amplitudes.

Amplitude ratio was introduced to get a more visual understanding of the comparison of the deviation of amplitude values between soundtriggered and amplitude-triggered measurements.

$A r a t i o = A p e a k A i n p u t$
(2)

<Fig. 16> shows the test results for the amplitude ratio from both triggering methods. In comparison to sound-triggered measurements, amplitude-triggered measurements have higher amplitude ratio for 0.1g input amplitude. This shows that sound-triggered measurements are considerably more accurate at lower input amplitude measurements. At higher input amplitudes, the amplitude ratio becomes smaller and closer to 1.0.

### 5.2. Analysis of error rate in frequency data

To analyze the accuracy of the experiment, the error rate (ψf) was calculated using equation (3)3). The natural frequency obtained from the experiment is given as Fnatural with the input frequency being Finput. The range of error values is 0.00~1.19% for the amplitude-triggered measurements and 0.00~1.22% for the sound-triggered measurements. All of the error values are below 3% error, with most of the high error values being due to low input amplitude measurements (0.01g and 0.05g). This shows that the MEMS sensor is more accurate in obtaining natural frequency results.

$ψ f = F n a t u r a l − F i n p u t F i n p u t × 100$
(3)

Lower error values are observed from the natural frequency results compared to the amplitude results.

To get a clearer image of the deviation of the obtained natural frequency values from the input frequency values, frequency ratio was introduced. The formula for frequency ratio is given by equation (4).

$F r a t i o = F n a t u r a l F i n p u t$
(4)

<Fig. 17> shows the test results for the frequency ratio between the input frequency and the frequency from the triggered measurements. It can be seen that the frequency ratio approaches 1.0 as the frequency and input amplitude increase.

## 6. Conclusion

In this experiment, the performance of the sound-triggered and amplitude-triggered measurements using MEMS sensors were compared by a series of shaking-table tests.

It can be concluded that sound-triggered measurements have a wider range of frequency and amplitude measurements as it could measure for lower frequencies (0.5Hz) and lower amplitudes (0.01g and 0.05g). In addition, amplitude-triggering performs better in simultaneous measurements at higher amplitudes (0.5g and 1.0g). This was seen by the reduction in time lag at higher amplitudes. Soundtriggering showed better simultaneous recording in comparison to amplitude triggering at lower frequency and amplitude values because devices could respond more easily to sound triggering than amplitude triggering.

It was seen that both of the triggering methods show similar error rates. The error values were considerably lower at higher input amplitudes and frequencies. All natural frequency error values were low (less than 3%) which shows accuracy in obtaining natural frequency using MEMS sensors.

It can be seen that the performances of both of the triggering methods are good with sound triggering being better at lower amplitudes and amplitude triggering performing better at higher amplitudes.

## Acknowledgements

This research was financially supported by the National Research Foundation of Korea (NRF-2016R1A2B2014064). The writers are grateful to the authorities for their support.

## Figure

Equipments used (APS 145 and Keysight 33210A)

Experimental setup (APS 400 with specimen used)

Time history for sound trigger at 0.5Hz for (a) 0.01g and (b) 0.05g

Power spectrum for sound trigger at 0.5Hz for (a) 0.01g and (b) 0.05g

Time history for sound trigger at 1.0Hz for (a) 0.01g and (b) 0.05g

Power spectrum for sound trigger at 1.0Hz for (a) 0.01g and (b) 0.05g

Time history for sound trigger at 1.0Hz for 0.1g

Time history for amplitude trigger at 1.0Hz for 0.1g

Power spectrum for sound trigger at 1.0Hz for 0.1g

Power spectrum for amplitude trigger at 1.0Hz for 0.1g

Time history for sound trigger at 5.0Hz for (a) 0.1g, (b) 0.5g and (c) 1.0g

Time history for amplitude trigger at 5.0Hz for (a) 0.1g, (b) 0.5g and (c) 1.0g

Power spectrum for sound trigger at 5.0Hz for (a) 0.1g, (b) 0.5g and (c) 1.0g

Power spectrum for amplitude trigger at 5.0Hz for (a) 0.1g, (b) 0.5g and (c) 1.0g

Peak amplitude-to-input amplitude ratio

Measured natural frequency-to-input frequency ratio

## Table

List of laboratory equipments8),9)

Characteristics of MEMS sensors used1),7),10)

Shaking table experiment overview

## Reference

1. Shrestha, A., Dang, J., & Wang, X., "Development of a smart‐device‐based vibration‐measurement system: Effectiveness examination and application cases to existing structure", Structural Control Health Monitoring, Vol.25, No.3, 2018
2. Feng, M., Fukuda, Y., Mizuta, M., & Ozer, E., "Citizen sensors for SHM: use of accelerometer data from smartphones", Sensors, Vol.15, No.2, pp.2980~2998, 2015
3. Mapungwana, S., Jung, Y. S., Lee, J. H., & Yoon, S. W., "Comparison Between Performance of a Wireless MEMS Sensor and an ICP Sensor in Shaking Table Tests", Journal of Korean Association for Spatial Structures, Vol.18, No.4, pp.49~59, 2018
4. Reilly, J., Dashti, S., Ervasti, M., Bray, J. D., Glaser, S. D., & Bayen, A. M., "Mobile Phones as Seismologic Sensors: Automating Data Extraction for the iShake System", IEEE Transactions on Automation Science and Engineering, Vol.10, No.2, pp.242~251, 2013
5. Vibration Version 2 [Software]. (2013). Diffraction Limited Design LLC. Retrieved from http://www.dld-llc.com