wireless vibration sensor position to acquire better...

5
AbstractStructural health monitoring system has become a new field in research and industry. Combination of multi discipline field has become an important factor. This research aim is to find out placement sensor effect in bridge behavior. This research is using a wireless vibration sensor made by ITS, this consist of accelerometer with 3 axis. Sensor and bridge behavior is basic of research. This research creates 21 scenarios to find out behavior of Laboratory Bridge. These scenarios include sensor placement, sensor position, load design and range between loads to sensor. Vibration data was whitened with independent component analysis and being analyzed for mode change in finite element. Research result show there was a change in bridge behavior according to scenarios above. Comparison between two or more combo shows that sensor placement in SHM system must be design and calculate. This means that vibration data must have correction according to their placement in structure. KeywordsAccelerometer, Bridge, Vibration, SHM I. INTRODUCTION TRUCTURAL health monitoring become important now days. This condition happens because of development in sensor detection and awareness to protect structure. Sensor development creates a condition of small sensor, less energy and easy to use. Meanwhile, after several accidents in structure for example Kutai bridge accident. Awareness to acquire bridge healthy and behavior become more than before. Three different field of study has united to create method, device and algorithm to achieve awareness above. They are electronic engineering field to design sensor, Mechanical engineering for device and Civil engineering for analyze. However until now research in this field is not done yet. Gopalakrisnan [1] defines four parts in structural health monitoring system to develop: 1. Base condition for structure health 2. Method for detection, Non destructive method Arie Febry is doctorate student in Civil Engineering and Design Faculty, Sepuluh November Institute of Technology, Surabaya, East Java, Indonesia (+628125127525; [email protected]). Priyo Suprobo is Lecturer in Civil Engineering and Design Faculty, Sepuluh November Institute of Technology, Surabaya, East Java, Indonesia (- ; [email protected]). Faimun is Lecturer in Civil Engineering and Design Faculty, Sepuluh November Institute of Technology, Surabaya, East Java, Indonesia (-; [email protected]). 3. Sensor type and development 4. Analyze of structure at end of detection One important sensor is accelerator sensor. This sensor aim is find out vibration from load. This sensor is gaining vibration data in term of accelerator (g) vs time (ms). Basic concept for this sensor is second Newton law. Accelerator is one of sensor, which is often using in bridge monitoring. Accelerator record vibration data from bridge and transfer to data recording system. Accelerator has three axis sensors, which are one, two or three axes. This data is called as dynamic function. Bridge is structure which is commonly become subject of monitoring research. This happens because bridge has a large dynamic load then building [2,3,4,5,6] Structural health monitoring system (SHMS) has been used in several bridge since 1996 [2]. Until this time several method has been propose. All of this method has different way to find health of structure. Frequency of structure and displacement has become limit for bridge behavior [7,8,9,10,11]. Thus, a lot of research is trying to find out base on both conditions. This research aim is to find out bridge health monitoring in laboratory. Bridge monitoring was made to find out behavior in several sensor placement scenarios. II. BASIC THEORY A. Accelerator An accelerometer is a sensing element. It is sensing an acceleration or velocity to time (m/s s). Accelerator is measuring in unit of g (gravitation). This sensor is sensing vibration, shock, tilt, impact or motion. Basic formula of acceleration is show in this formula below: (1) There is various type of accelerometer can be use. It is depend on what kind of result to find. However, There are five factors to consider to choose accelerator sensor. These five factor will influence recorded vibration data. 1. Dynamic range This factor is controlling maximum or minimum of amplitude of sensor. It is define as g factor. Wireless Vibration Sensor Position to Acquire Better Bridge Healthy Monitoring in Laboratory Scale Arie Febry, Priyo Suprobo, and Faimun S 2nd International Conference on Innovations in Engineering and Technology (ICCET’2014) Sept. 19-20, 2014 Penang (Malaysia) http://dx.doi.org/10.15242/IIE.E0914027 72

Upload: buikiet

Post on 16-Mar-2019

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Wireless Vibration Sensor Position to Acquire Better ...iieng.org/images/proceedings_pdf/1576E0914027.pdf · disturbing data from test. Several methods are use for this task ... Grafik

Abstract—Structural health monitoring system has become a new

field in research and industry. Combination of multi discipline field

has become an important factor. This research aim is to find out

placement sensor effect in bridge behavior.

This research is using a wireless vibration sensor made by ITS,

this consist of accelerometer with 3 axis. Sensor and bridge behavior

is basic of research. This research creates 21 scenarios to find out

behavior of Laboratory Bridge. These scenarios include sensor

placement, sensor position, load design and range between loads to

sensor. Vibration data was whitened with independent component

analysis and being analyzed for mode change in finite element.

Research result show there was a change in bridge behavior

according to scenarios above. Comparison between two or more

combo shows that sensor placement in SHM system must be design

and calculate. This means that vibration data must have correction

according to their placement in structure.

Keywords—Accelerometer, Bridge, Vibration, SHM

I. INTRODUCTION

TRUCTURAL health monitoring become important now

days. This condition happens because of development in

sensor detection and awareness to protect structure. Sensor

development creates a condition of small sensor, less energy

and easy to use. Meanwhile, after several accidents in

structure for example Kutai bridge accident. Awareness to

acquire bridge healthy and behavior become more than before.

Three different field of study has united to create method,

device and algorithm to achieve awareness above. They are

electronic engineering field to design sensor, Mechanical

engineering for device and Civil engineering for analyze.

However until now research in this field is not done yet.

Gopalakrisnan [1] defines four parts in structural health

monitoring system to develop:

1. Base condition for structure health

2. Method for detection, Non destructive method

Arie Febry is doctorate student in Civil Engineering and Design Faculty,

Sepuluh November Institute of Technology, Surabaya, East Java, Indonesia

(+628125127525; [email protected]). Priyo Suprobo is Lecturer in Civil Engineering and Design Faculty,

Sepuluh November Institute of Technology, Surabaya, East Java, Indonesia (-

; [email protected]). Faimun is Lecturer in Civil Engineering and Design Faculty, Sepuluh

November Institute of Technology, Surabaya, East Java, Indonesia (-;

[email protected]).

3. Sensor type and development

4. Analyze of structure at end of detection

One important sensor is accelerator sensor. This sensor aim

is find out vibration from load. This sensor is gaining vibration

data in term of accelerator (g) vs time (ms). Basic concept for

this sensor is second Newton law.

Accelerator is one of sensor, which is often using in bridge

monitoring. Accelerator record vibration data from bridge and

transfer to data recording system. Accelerator has three axis

sensors, which are one, two or three axes. This data is called

as dynamic function.

Bridge is structure which is commonly become subject of

monitoring research. This happens because bridge has a large

dynamic load then building [2,3,4,5,6]

Structural health monitoring system (SHMS) has been used

in several bridge since 1996 [2]. Until this time several

method has been propose. All of this method has different way

to find health of structure.

Frequency of structure and displacement has become limit

for bridge behavior [7,8,9,10,11]. Thus, a lot of research is

trying to find out base on both conditions.

This research aim is to find out bridge health monitoring in

laboratory. Bridge monitoring was made to find out behavior

in several sensor placement scenarios.

II. BASIC THEORY

A. Accelerator

An accelerometer is a sensing element. It is sensing an

acceleration or velocity to time (m/s – s). Accelerator is

measuring in unit of g (gravitation). This sensor is sensing

vibration, shock, tilt, impact or motion. Basic formula of

acceleration is show in this formula below:

(1)

There is various type of accelerometer can be use. It is

depend on what kind of result to find. However, There are five

factors to consider to choose accelerator sensor. These five

factor will influence recorded vibration data.

1. Dynamic range

This factor is controlling maximum or minimum of

amplitude of sensor. It is define as g factor.

Wireless Vibration Sensor Position to Acquire

Better Bridge Healthy Monitoring in Laboratory

Scale

Arie Febry, Priyo Suprobo, and Faimun

S

2nd International Conference on Innovations in Engineering and Technology (ICCET’2014) Sept. 19-20, 2014 Penang (Malaysia)

http://dx.doi.org/10.15242/IIE.E0914027 72

Page 2: Wireless Vibration Sensor Position to Acquire Better ...iieng.org/images/proceedings_pdf/1576E0914027.pdf · disturbing data from test. Several methods are use for this task ... Grafik

2. Sensitivity

This factor is controlling output signal per input. This

means ability to detect motion.

3. Frequency response

This factor is frequency range for which the sensor might

detect motion.

4. Sensitive axis

Acceleration sensor can be detecting from one to three

axes.

5. Size and mass

Rule of this factor is mass of accelerometer must be less

then mass of system to be monitor.

Calibration of sensor is using:

a. Linearity of sensor, which is consider as maximum

deviation from straight line.

( ) (2)

b. Sensitivity of sensor

(3)

Accelerometer result is vibration data. It is velocity to time.

Figure 2.1 shows vibration data as below

Fig. 1 Sample of data recorded using sensor

B. Vibration data Filter

Data filter function is cleaning data from un-use or

disturbing data from test. Several methods are use for this

task. Mostly in structural health monitoring (SHM) is using

independent component analysis (ICA) and it descent

(FastICA), (pnGICA) as propose in several research

[12,13,14,15,16].

Independent component analysis (ICA) method is a

technique to find linier independent component using statistic.

It is opposite from principal component analysis. ICA does not

use principal component as variance.

Structural health monitoring uses this method for cleaning

vibration data or returning position of vibration wave into zero

Y-axis as default. ICA method for vibration data is proposed

to be use in linier and must use for independent component.

Concept of ICA is described at equation 4.

( ) ∑ ( ) (4)

FastICA is newest method from ICA which is more

efficient for use. FastICA is based on fixed point iteration

scheme for finding non gaussianity of wT x, and can also

derived as an approximate Newton iteration [3]. The basic

form of the fastICA algorithm is as follow:

1. Choose an initial weight vector w

2. Let wT = E(xg(w

Tx)) - E(g

2(w

Tx))w

3. Let w = wT / w

T

4. If not converged, go back to step 2

C. Finite Element

Finite element package was used to find bridge behavior.

Time history load was used as load for structure model.

III. RESEARCH METHODOLOGY

A. Sensor device

A sensor device was built to acquire data. This device

already compare with 3DM vibration sensor. Calibration has

been done to this device. Figure 2 show device for vibration

sensor.

Fig. 2 Developed Wireless Vibration Sensor

Sensor in this research is wireless sensor. Sensor device aim

is to detect vibration by using accelerometer. It will be an

accelerator sensor using tri-axial axis. Sensor can be use is two

legs sensor. It means that beside accelerometer other sensor

can be use in it.

Frequency of vibration sensor can be adjusted into three

conditions. Three level of frequency can be used. This will be

use in field and laboratory test correction.

B. Bridge Model

Laboratory scale for bridge model is shows in figure 3 to 5

below:

Fig. 3 Bridge Model for laboratory scale

0 20 40 60 80 100 120 140 160 180 200-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

sampling

nila

i get

aran

G(m

/s2)

Grafik sensor acceleromter secara real time

accelerometer bidang x

accelerometer bidang y

accelerometer bidang z

X Axis

Y Axis

Z Axis

2nd International Conference on Innovations in Engineering and Technology (ICCET’2014) Sept. 19-20, 2014 Penang (Malaysia)

http://dx.doi.org/10.15242/IIE.E0914027 73

Page 3: Wireless Vibration Sensor Position to Acquire Better ...iieng.org/images/proceedings_pdf/1576E0914027.pdf · disturbing data from test. Several methods are use for this task ... Grafik

Fig. 4 Restraint in bridge laboratory scale

Fig. 5 Connection type in bridge model

Laboratory model specifications are:

Girder bridge use WF 150.150

Yield point is 240 Mpa

C. Load Type

Impact load was used to find vibration data. Impact load

was made by using rubber hammer with specific load and

distance for falling.

In this research, rubber hammer were 585 and 810 grams

with several type of falling distance.

D. Detection Scenario

Scenario was made with several combinations which were

position of sensor, weight of hammer, sensor position with

load as define in table 1.

TABLE 1

SCENARIO FOR SENSOR IN FLANGE POSITION

Combo Location Distance

(cm)

Height

(cm)

Weight

(gram)

1 Middle 20 20 565

2 Middle 20 20 810

3 Middle 150 20 565

4 Middle 150 20 810

5 Middle 150 40 565

6 Middle 240 20 565

7 Middle 240 20 810

8 End 150 20 565

9 End 150 40 565

10 End 240 20 565

11 End 240 40 565

Combo Location Distance

(cm)

Height

(cm)

Weight

(gram)

12 End 150 60 565

13 End 20 20 565

14 End 150 60 565

and for web position is show in table 2.

TABLE II

SCENARIO FOR SENSOR IN WEB POSITION

Combo Location Distance

(cm)

Height

(cm)

Weight

(gram)

15 End 20 20 565

16 End 150 20 565

17 End 150 40 565

18 End 150 60 565

19 End 240 20 565

20 End 240 40 565

21 End 240 60 565

IV. RESULT & DISCUSSION

A. Vibration data

Based on 21 scenarios, vibration data is taken. Vibration

data was cleaning and whitening. Figure 6 to 8 show data after

this process has been done.

Fig. 6 Whitten Vibration Data in X axis (Combo1 )

Fig. 7 Whitten Vibration Data in Y axis (Combo1 )

-6.000

-4.000

-2.000

0.000

2.000

4.000

6.000

1

16

31

46

61

76

91

106

121

136

151

166

-2.000

-1.000

0.000

1.000

2.000

1

15

29

43

57

71

85

99

113

127

141

155

169

2nd International Conference on Innovations in Engineering and Technology (ICCET’2014) Sept. 19-20, 2014 Penang (Malaysia)

http://dx.doi.org/10.15242/IIE.E0914027 74

Page 4: Wireless Vibration Sensor Position to Acquire Better ...iieng.org/images/proceedings_pdf/1576E0914027.pdf · disturbing data from test. Several methods are use for this task ... Grafik

Fig.8 Whitten Vibration Data in Z axis (Combo1 )

This package data above is taken from combo 1. Data is

splitting into different axes, then by using FastICA method

data is filtered. After that data is whiten and will be use as

time history in finite element.

B. Bridge Behavior

Finite element method is using in this analysis to find out

bridge behavior. Models are creating and analyze to find out

mode of bridge for their bridge standard.

Basic consideration in bridge healthy was mode change

cause by time history [17,18]. 21 scenarios was made to

analyze bridge healthy. Result from 21 scenarios would be

compare and taken several condition that can result in this

research aim.

Fig. 9 First Mode from combo one

Fig. 10 First mode from combo two

Figure 9 and 10 show different mode behavior. Combo 1

and combo 2 were different in weight of load. These confirm

the relation between bridges healthy with load. This

comparison show effect of load in bridge behavior.

Fig. 11 Second mode from combo one

Fig. 12 Second mode from combo six

Figure 11 and 12 show that second mode is more use full to

gain information in bridge behavior. Combo 6 is different with

combo 1 in load position. Second mode in combo six shows a

deflection in middle girder, meanwhile from combo 1 show

deflection at the end of beam. This different behavior can be

defined for sensor calibration.

Base on data above, several conditions have been taken as in

table 3

TABLE 3

COMPARRISON IN BRIDGE BEHAVIOR SCENARIO

Combo Translation Z Rotation Y

01 : S20.20.565T 0.00123 0.000200

02 : S20.20.810T 0.00166 0.000325

03 : S150.20.565T 0.00040 0.000082

05 : S150.40.565T 0.00003 0.000006

06 : S240.20.565T 0.00052 0.000120

08 : S150.20.565U 0.0009 0.00019

15 : B20.20.565U 0.00018 0.000036

Base on data above sensor position, working load and

position of load create a different behavior and deviation from

laboratory model. Compare different condition from data

above show result:

a. Weight load different (Combo 1 with Combo 2)

According to bridge behavior, more weight create a

more translation and rotation by adding 35% more

weight in load create a change in translation by 35% and

rotation by 62.5%

b. Sensor position in Bridge Body (Combo 1 with Combo

15)

Deviation happens in sensor position, this create a

different behavior in bridge model. Sensor in web creates

a derivation in translation about 85% and in rotation

about 82%.

c. Sensor Position According to Load (Combo 1 : Combo 3

: Combo 6)

Result shows sensor that was placed in front of load

creates better than other position in translation or in

rotation.

d. Height of load falling (Combo 3 and Combo 5)

Height of load falling show different result in rotation

more height mean more rotation value. Nevertheless, this

does not happen in translation value.

-6.000

-4.000

-2.000

0.000

2.000

4.000

6.000

1

14

27

40

53

66

79

92

105

118

131

144

157

170

2nd International Conference on Innovations in Engineering and Technology (ICCET’2014) Sept. 19-20, 2014 Penang (Malaysia)

http://dx.doi.org/10.15242/IIE.E0914027 75

Page 5: Wireless Vibration Sensor Position to Acquire Better ...iieng.org/images/proceedings_pdf/1576E0914027.pdf · disturbing data from test. Several methods are use for this task ... Grafik

V. CONCLUSION

Bridge health behavior shows some derivation in sensor

position. Result from this research can be used to achieve

health of bridge and pattern of sensor placement in field. This

research shows conclusion in research as below:

1. Bridge behavior is connected with weight in

structure. This means that design of largest vehicle

must be accomplished in real condition.

2. Direct line with vehicle location is the best position

for sensor.

3. Placement of sensor in middle of bridge show most

accurate data, by position data read of vehicle from

end of bridge.

4. Shock load in bridge create a large different in bridge

rotation.

5. Position for sensor must be in the right place to create

an accuracy and valuable data to predict bridge

health.

ACKNOWLEDGMENT

Ministry of Education and Culture, Indonesia sponsor this

research. Authors really appreciate for their fund and help.

And also for ITS LPPM for their support in this research

This research is part of Indonesia Structural Health

Monitoring System (I-SHMS) which is still going to develop

integrated SHM system

REFERENCES

[1] Gopalakrishnan, S, Ruzzene, M and Hanagut, S, (2011), Computational

Techniques for Structural Health Monitoring, 1st edition , Springer, London

http://dx.doi.org/10.1007/978-0-85729-284-1

[2] Chen G. (2012). “Structural Health Monitoring in Transportation Infrastructure Application – New Persfective”. Sensing Technologies

for Transportation Application Workshop at TRB 91st Annual Meeting.

22 January 2012. Washington DC. [3] Ettouney, M and Alampalli, S, (2012), Infrastructure Health in Civil

Engineering Volume 1, 1st Edition, CRC Press, Boca Baton.

[4] Papadioti D, Papadimitriou C and Panagiotis P (2011). “Bridge Health Monitoring Techniques: Integrating Vibration Measurements and

Physics-based Models”. IBSBS. 2011. Athens. Greece

[5] Setijadi, E. Suwadi. Et all (2013). “Design of Large Scale Structural Health Monitoring System for Long Span Bridges Based On Wireless

Sensor Network”. ICAST Umedia 2013. 2-4 November 2013. Aizu

Wakamatsu. [6] Zhu, D. Wang Y and Brownjhon J. (2012). “Vibration Testing of a

Steel Girder Bridge using Cabled and Wireless Sensors”. Frontier Architecture and Civil Engineering China. Volume 5(3), 2011. Pages

249-258.

http://dx.doi.org/10.1007/s11709-011-0113-y [7] Wenzel, H, (2009), Health Monitoring of Bridges, 1st Edition, John

Wiley & sons, Ltd, West Sussex.

http://dx.doi.org/10.1002/9780470740170 [8] Brownjohn J. Magalhes F. Et all (2009). “Ambient Vibration Re-

Resting of The Humber Bridge”. 4th International Conference on

Structural health Monitoring of Intelligent Infrastructure (SHMII). 22 – 24 July 2009. Zurich.

[9] McCullagh J, Peterson R.L. Et all (2012). “Short Term and Long Term

Testing of A Vibration Harvesting System For Bridge Health Monitoring”. PowerMEMS. 2-5 December 2012. Atlanta.

[10] Oth, A. and Picozzi M. (2012). “Structural Health Monitoring Using

Wireless Technologies: An ambient Vibration Test on the Adolphe Bridge, Luxenbourg City”. Advances in Civil Engineering. Volume

2012 Article ID 876174. Pages 17.

http://dx.doi.org/10.1155/2012/876174

[11] Resnik B. and Ribakov Y (2012). “Impelementation and Analysis of

Vibration Measurement Obtained from Monitoring The Magdeburg Water Bridge”. 7th International Conference on Material Technologies

and Modelling. Israel

[12] Cheng J. Nakayama M. Et all (2008). “Bridge Diagnosis System By Using Wireless Sensor Network and Independent Component

Analysis”. SICE Annual Conference 2008. 20–22 August 2008. Japan.

[13] Hyvarinen A and Oja Erkki (2002). “Independent Component Analysis: Algorithms and Applications”. Journal of Neural Networks. Volume

13(4-5), 2002. Pages 411-430

http://dx.doi.org/10.1016/S0893-6080(00)00026-5 [14] McNeill S. I. (2007). “Use of ICA techniques to Decompose Free

Vibration Data”. IMAC-XXV: Conference & Exposition on Structural

Dynamics, 2007. Houston. [15] Yokote R. and Matsuyama Y (2012). “Rapid Algorithm for

Independent Component Analysis”. Journal of Signal and Information

Processing. Volume 3, 2012. Pages 275-285 http://dx.doi.org/10.4236/jsip.2012.33037

[16] Zheng J. Yeh Y. and Ogai H. (2011). “Bridge Diagnosis by Using Non

Linier Independent Component Analysis and Displacement Analysis”. SICE Journal of Control, Measurement, and System Integration.

Volume 4, 2011. Pages 315-321.

http://dx.doi.org/10.9746/jcmsi.4.315 [17] Arredondo M. A. Tibaduiza D. A. and Mujica L. E., et al (2014).

“Data Driven Multivariate Algorithms for Damage detection and

Identification: Evaluation and Comparison”. Structural Health Monitoring Journal. Volume 13(1), 2014. Pages 19-32

http://dx.doi.org/10.1177/1475921713498530 [18] Whelan M.J. Gangone M.V.. Et all (2008). “Wireless Vibration

Monitoring for Damage Detection of Highway Bridges”. SPIE Smart

Structures Symposium. 2008. San Diego.

2nd International Conference on Innovations in Engineering and Technology (ICCET’2014) Sept. 19-20, 2014 Penang (Malaysia)

http://dx.doi.org/10.15242/IIE.E0914027 76