Operational Modal Analysis of Historical Constructions using Commercial

Wireless Platforms

Luis F. Ramos(*), Rafael Aguilar, and Paulo B. Loureno

Department of Civil Engineering ISISE, University of Minho

Campus de Azurm 4800 - 058 Guimares, Portugal

(*) Tel. +351 253510200, Fax +351 253510217, e-mail: lramos@civil.uminho.pt

ABSTRACT

Operational modal analysis is currently applied in structural dynamic monitoring studies

using conventional wired based sensors and data acquisition platforms. However, this

approach becomes inadequate in cases in which aesthetic concerns arise (e.g. in cultural

heritage buildings) or in which the use of wires greatly impacts the monitoring system

cost and creates difficulties in the maintenance and deployment of the data acquisition

platforms. In these cases, the use of a (Wireless Sensor Network) WSN and (Micro-

Electro-Mechanical Systems) MEMS would clearly benefit the applications. This work

illustrates an attempt to apply the wireless technology for the structural health monitoring

of historical masonry constructions in the context of operational modal analysis.

The paper presents the state of the art of the wireless monitoring systems making a review

of the applications in the civil engineering field. Subsequently, commercial wireless

based platforms and conventional wired based systems are applied to study one laboratory

specimen and one structural element from a 15th Century building located in Portugal.

The results achieved in this study showed that, in comparison to the conventional wired

sensors, the commercial wireless platforms have poor performance with respect to the

acceleration time series recorded and the detection of modal shapes. However, reliable

results were obtained for the measured frequencies.

KEYWORDS

Operational modal analysis, Wireless sensors networks, Micro-Electro-Mechanical

systems, Conventional wired based systems, Wireless based systems.

NOTATION

Natural frequency of the system h Hankel matrix

Damping coefficient j Number of columns in the

Hankel Matrix

k Process noise Yf Future information of the

Hankel matrix

vk Measured noise Yfref Past information of the Hankel

matrix

xk Discrete-time space state vector L,Q Factors from a LQ factorization

yk Observation vector Eigenvalues matrix

A Discrete-time state matrix S Observability matrix

B Discrete-time input matrix Complex modes shape matrix

C Discrete-time output matrix Eigenvectors matrix

D Discrete-time direct transmission

matrix

Real eigenvalues or real

eigenfrequencies

1. INTRODUCTION

The conservation of cultural heritage buildings, which provide identity to a region or

country, is a key aspect to modern societies, given the societal and economical demands.

This work aims at evaluating possible damages and performing structural health

monitoring taking into account the modern framework of minimum repair and the use of

non intrusive methods. Advancements on Micro-Electro-Mechanical Systems (MEMS)

and the wide range of alternatives on wireless communications are turning Wireless

Sensor Network (WSN) technology into promising candidates to adopt new structural

monitoring solutions for this field. The objective of the present work is to adopt

commercially available WSN and MEMS technologies in Operational Modal Analysis

(OMA) case studies to evaluate their capabilities and provide future research needs.

The paper starts with a general description of conventional wired based systems and the

state of the art of wireless based sensors and Data Acquisition (DAQ) equipments used

for structural dynamic monitoring. The most important methods for processing OMA data

are presented with an emphasis in the Stochastic Subspace Identification (SSI) method,

which is used to process the experimental results. Finally, two cases studies using

commercial wireless platforms for OMA are shown and compared to conventional

systems. It is noted that one of the applications is to a masonry chimney, which is rather

novel, as almost all existing references to WSN and MEMS are focused on bridges.

2. OVERVIEW OF DYNAMIC MONITORING SYSTEMS FOR CIVIL

ENGINEERING STRUCTURES

In the recent years, numerous applications of modal analysis covering wide areas of the

engineering have been reported in the literature. In the civil engineering field, modal

analysis is used to calibrate analytical models, to provide a better understanding of the

behaviour of the structures, to control quality of execution or to perform damage

detection.

Currently, significant hardware developments have also occurred in the structural

monitoring field. The sensors used for these applications involve significant wiring (fibre-

optic cables or other physical transmission medium) and centralized data acquisition

systems with remote connections.

The fact that conventional sensors are wired might lead to high installation costs,

problems in maintenance and severe difficulties in placing the sensors in selected

locations. Therefore, the recent years have witnessed an increasing interest in a new

technology based on the new technologies as low-cost alternatives for monitoring [1].

2.1 Wired Based Systems

Wired based systems (also called here conventional systems) are composed by three parts:

1) measuring sensors; 2) DAQ equipments; and, in some cases, 3) remote connection

systems. Measuring sensors are connected with cables to the DAQ equipments, which

can be remotely connected to a central system.

Conventional measuring sensors used for dynamic identification tests are generally

piezoelectric, piezoresistive, capacitive or force-balanced accelerometers. For data

acquisition purposes, platforms with capability of moderate sampling rates (from 100 Hz

to 2000 Hz) and Analog Digital Converters (ADCs) with resolutions higher than 16 bits

are usually chosen. In the case of remote connection systems the most popular approaches

use IEEE 802.11a, b, and g standards or cellular data (such as CDMA, GSM/GPRS or

EDGE) for communication purposes [2].

2.2 Wireless Based Systems

The research efforts in many scientific areas, such as physics, microelectronics, control,

material science, etc., are oriented to the creation of smaller, autonomous and easier to

handle mechanisms for sensing purposes. In the area of measuring physical parameters,

these goals were successfully achieved via the integration of MEMS with low power and

high frequency transceivers, joined in silicon chips. Sensor prototypes, called motes,

were developed to reach four attributes: sensing, processing, communication and

actuation.

A mote is an autonomous, compact device, and a sensor unit that has the capability of

processing and communicating wirelessly. One of the biggest strengths of motes is that

they can form networks, known as Wireless Sensor Networks (WSN), which allows the

units to co-operate between themselves.

Wireless technology is being used for a wide range of applications such as military,

environmental monitoring (e.g. indoor for emergency services or outdoor for agriculture

applications), support for logistics (e.g. considering the use of wearable motes in firemen),

human centric (motes for health science and health care) and robotics. For more details,

see Arampatzis et al. [3].

The use of wireless technology with embedded MEMS for structural monitoring was first

proposed by Straser and Kiremidjian [4-7], aiming at the integration of wireless

communications with sensors in order to develop a near real time monitoring system.

After these preliminary studies, many efforts to improve the technology had been carried

out.

The first commercial wireless platform with embedded MEMS was developed by the

University of California-Berkeley [8] and subsequently commercialized by Crossbow [9]

in 1999. Figure 1 illustrates updated information about the state of the art, based on [8].

Figure 1 - State of the art of the wireless technology for structural monitoring

A monitoring system based on WSN platforms with embedded MEMS is composed by

three parts: 1) Measuring units, 2) base station and, in some cases, 3) remote connection

system. The equipments and the technology used for the last part of the system (remote

Straser and Kiremijdian, (1996)

Bennett et al.,

(1999)Lynch et al.,

(2001, 2002a,

2002b)Mitchell et al.,

(2002)

Kottapalli et al.,(2003)

Lynch et al.,

(2003, 2004a, 2004b)

Aoki et al., (2003)

Basheer et al., (2003)

Casciati et al., (2003, 2004)

Wang et al., (2003)

Wang et al., (2004)

Gu et al.,(2004)

Mastroleon et al., (2004)Shinozuka, (2003)

Chung et al.,(2004)Ou et al.,(2004)

Sazanov et al.,(2004)

Farrar et al.,(2005)

Allen, (2005)

Wang et al.,(2005) Pei et al.,(2005)

[1999]

UC

Berkeley

Crossbow

WEC

[2000]

UC Berkeley

Crossbow

Rene

[2002]

UC Berkeley

Crossbow

MICA[2003]

UC Berkeley

Crossbow MICA2

Intel iMote (Kling, 2003)

Microstrain (Galbreath et al.,2003)

2000

2002

2004

2006

: Academic prototype

: Commercial prototype

Lynch, (2007)

2008

Straser and Kiremijdian, (1998)Kiremijdian et al., (1997)

Straser et al., (1998)

connection) are, usually, the same as for conventional systems. Therefore, attention is

given here to the other parts.

2.2.1 Wireless Measurement Units

A wireless measurement unit can be understood as three functional subsystems working

in parallel: sensing interface (MEMS act as sensors and DAQ subsystems), computational

core (microcontrollers and memory) and a wireless communication module (wireless

radio to transmit or receive data).

MEMS is an emerging technology through which miniature mechanical systems are built

making use of the standard Integrated Circuits (IC) technologies on the same chip as the

electronic circuitry. The main advantage of MEMS is that, because of the effectiveness

in their fabrication process, they can perform measurements at relatively low cost and low

power consumption.

The field of MEMS has been developed since the end of the 1980s while, in comparison

silicon-based sensors and actuators go back to the 1970s [10]. Currently, MEMS are used

for many applications such as communications (mobile phones), finding industrial,

automotive industrial, medical and security purposes [11]. For dynamic monitoring of

civil engineering structures, mechanical microsensors (microaccelerometers) are the most

appropriate.

Microaccelerometers are built on a variety of principles like capacitance, strain and

piezoelectricity. Commercial accelerometers are primarily based on the capacitive

principle and are able to perform measurements in one, two or three axis. The

measurement range of these sensors is between 2 g to 400 g and the sensitivity range

vary about 150 mV/g to 2000 mV/g. Due to the fact that in dynamic structural monitoring

very low vibrations are measured, microaccelerometers with a small range and high

sensitivity should be chosen.

If the wireless based monitoring platforms are considered as alternatives to conventional

systems, the DAQ systems embedded in such platforms should be able to collect sensor

data with equivalent accuracy. For dynamic structural monitoring purposes, the sensing

interface should be therefore capable of recording at relatively high sample rates (up to

200 Hz) and the ADCs should have resolutions higher than 16 bits. As it is shown in

Lynch and Loh [8], commercial platforms have been only equipped so far with lower

resolutions ADC.

The computational core is responsible for the operation of the wireless sensing unit,

including data collection, implementation of algorithms for data processing and managing

the flow of data through the wireless communication channel [12]. The computational

core is composed by microcontrollers assembled with on-chip computing resources with

enough memory to store the recorded data and the embedded computing software [13].

A broad assortment of microcontrollers is commercially available such as the

ATmega103L (mica platforms), Atmel ATmega128L (mica2 platforms) and the

ARM7TDMI (iMote2 platforms). The range of memory available in the platforms varies

from 4 kB to 64 kB for the RAM memory and from 128 kB to 512 kB for the flash

memory. The size of the bus varies from 8 bits to 32 bits and the speed of the clock also

varies from 4 MHz to 12 MHz.

The wireless communication module provides an interface for the exchange of data with

other wireless units or with the base station. Important considerations like communication

reliability, communication range, allowed frequency allocation, and data transfer rate

(factors related to the communication standard) should be considered. Due to the fact that

the wireless modules are the most power consuming components in the measurement unit,

the power consumption aspect should also be taken into account [14].

Currently, commercial platforms are not yet implemented with communication standards

and have data transmission rates varying from 40 kb/s (micas platforms) to 250 kb/s

(imote2 platform).

2.2.2 Base Station

A base station (also known as gateway) is a receiver/transmitter unit that serves as the

connection point between the WSN and the computer where the data are collected. The

base station is composed by a wireless communication module coupled with an interface

board in charge of data collecting and mote programming.

The characteristics of the wireless communication module are the same as the

characteristics of the modules used in the measuring units. The power consumption in

this case is not a critical issue, as the wireless module is directly connected to the

computer.

The interface board is mainly connected to a computer though a serial port RS-232. Other

physical mediums, like the USB port or the JTAG interface, can also be used with the

same purpose. In the case of the serial port RS-232, the transmission rates vary from

20 kb/s to 115.2 kb/s. With the use of USB ports, faster data transmission rates like

1.5 Mb/s (USB 1.0), 480 Mb/s (USB 2.0) or 5 Gb/s (USB 3.0) can be achieved.

2.2.3 Operating System of the Motes

The operating system provides an abstraction of the machine hardware and is in charge

of reacting to events and handling access to memory, CPU, and hardware peripherals. In

memory constrained hardware devices like those of sensor boards, the effectiveness in

the operating system largely affects the response in the target application [15].

The wireless based commercial platforms available at the market use TinyOS as operating

system. TinyOS is a free and open source component-based operating system developed

by the University of California of Berkeley in co-operation with Intel which, in its first

release, was presented on 1999 [16].

TinyOS utilizes a unique software architecture specifically designed for the severe

constraints of the sensor network nodes [17]. The components in TinyOS are written in a

Network Embedded Systems C (nesC), a dialect of C that adds some new features to

support the structure and execution model [18]. The supplemental tools that the system

uses come mainly in the form of Java and shell script front-ends.

2.2.4 Applications of the Wireless Based Systems to Civil Engineering Structures

The first case of study in which wireless based systems were used to monitor civil

engineering structures, was the Alamosa Canyon Bridge in 1998 [6]. After that test, more

bridges were considered [19-24]. However, the only use in dynamic monitoring of large

buildings seems to be 79 stories Di Wang Tower in China [25]. Due to fact that masonry

structures are difficult to excite and due to the low resolutions capabilities of the

commercial MEMS, only one application was found, namely the Aquila Tower in Italy

[26].

3. OPERATIONAL MODAL ANALYSIS OF CIVIL ENGINEERING

STRUCTURES

To carry out structural dynamic monitoring, two different groups of techniques can be

used, the Input-Output and the Output-Only techniques.

The Input-Output techniques are based on the estimation of a set of Frequency Response

Functions (FRFs) relating an applied force to the corresponding response at several points

along the structure. Equipments like impulse hammers, impulse excitation devices,

electro-dynamic shakers, eccentric mass vibrator and servo-hydraulic shakers are

commonly used. The main drawbacks of those equipments are the relatively low spectral

frequency resolution estimations and the lack of energy to excite some relevant modes of

vibration [27].

On the other hand, Output-Only methods (also known as OMA) are based on the premise

that wind, traffic and human activities can adequately excite structures. The main

assumption of OMA is that the ambient excitation input is as a Gaussian white noise

stochastic process in a frequency range of interest.

Due to the nature of the excitation, the response includes not only the modal contributions

of the ambient forces and the structural system, but also the contribution of the noise

signals from undesired sources. In this way the measurements reflect the response from

the structural system and also from the ambient influence; and therefore the identification

techniques must have the ability to separate them.

As it is summarized in Cunha et al. [27], the methods to process OMA results are divided

in two groups: nonparametric methods, essentially developed in frequency domain,

(Group G1 in Figure 2) and parametric methods, developed in time domain (Group G2 in

Figure 2).

Figure 2 Methods to process OMA results (adapted from [28])

Considering previous experience, when dynamic monitoring is performed, there is no best

method. The Stochastic Subspace Identification method (SSI) and its variants give usually

the most reliable results, being adopted in this work.

The SSI method was originally proposed by Van Overschee and De Moor [29] and then

modified by Peeters and De Roeck [30], as the so-called Data-Driven Stochastic Subspace

Identification method (SSI-Data).

The SSI-Data method is based on the stochastic space model theory from output-only

measurements and is focused in the identification of the state matrix A and the output

matrix C, which contains the modal information of the studied system. This method uses

robust numerical techniques such as QR-factorization, singular value decomposition

(SVD) and least squares. The QR-factorization results in a significant data reduction,

Response

time series

y (t)

Welch method

Random Decrement

(RD) method

Estimates of

RD functions

Dy(t)

Direct Method

FFT based Method

Estimates of

Power Spectral

Density

Functions

Sy(f)

Estimates of

Correlation

Functions

Ry(f)

Data-Driven Stochastic Subspace Identification (SSI DATA) method

Peak Picking (PP) method

FDD and EFDD method

RD PP method

RD FDD and RD EFDD methods

ITD and MRITD methods

LSCE and PTD methods

SSI-COV method

Modal

Parameters

fi

i

fi

FFT

FFT

FFT

SVD

SVD

LS, EVD

LS, EVD

SVD, LS, EVD

QR, SVD, LS, EVD

Numerical techniques used:

FFT : Fast Fourier transform

SVD : Singular Value decomposition

LS : Least Squares fitting

EVD : Eigenvector decomposition

QR : Orthogonal decomposition

PolyMAX method

whereas the SVD is used to eliminate the system noise. A summary of the method is

presented in Figure 3.

Figure 3 Flow chart of the SSI-Data method

4. OPERATIONAL MODAL ANALYSIS USING COMMERCIAL WIRELESS

BASED SYSTEMS

4.1 Measurement Sensors and Data Acquisition Equipments

As discussed above, conventional wired based equipments were successfully used in the

past for structural monitoring, being here considered as reference for comparison

purposes.

The conventional wired based sensors used were the high sensitivity piezoelectric

accelerometers model PCB 393B12 [31]. For DAQ purposes, the NI-USB9233 [32] board

with an ADC resolution of 24 bits was selected.

In the case of the wireless based systems, the Crossbow technology [9] was chosen, as it

offers inexpensive solutions with low powering boards and platforms with embedded

microaccelerometers.

kkkk wBuAxx 1

kkkk vDuCxy

kkk wAxx 1

kkk vCxy

General solution for Deterministic-

Stochastic State-Space models

x: state vector

y: observation vector

Solution when

Output-Only

methods are used

future""

past""

...

............

...

...

...

............

...

...

1

22212

21

11

21

21

110

f

ref

p

jiii

Niii

jiii

ref

ji

ref

i

ref

i

ref

j

refref

ref

j

refref

Y

Y

yyy

yyy

yyy

yyy

yyy

yyy

jh

H

H

H

H

H

Q

Q

Q

Q

LLLL

LLL

LL

L

LQh

4

3

2

1

44434241

333231

2221

11

0

00

000

H

LLL VU

L

L

L

41

31

21

2

1

2

1

210

0

V

V

S

SUUVU HLLL

21

11SUOi

00

00

00

0

0

41

31

21

1

333231

2121

42411

L

L

L

O

LLL

LL

LLO

C

Ai

i

1A

C

)Re(

Estimation of

modal parameters

LQ factorization

of h.

Construction

of Hankel

matrix

SVD of a part of

the h matrix

Selection of

the system

order n to

split the

singular

values

Estimation

of the A

and C

matrixes

Extend

observability

matrix

1

2

3

4

5

6

7

8

9

For comparison purposes Table 1 presents the characteristics of the microaccelerometer

ADXL202 embedded in the Crossbow platforms and the piezoelectric accelerometer PCB

393B12.

Table 1 Characteristics of a MEMS and a conventional piezoelectric accelerometer.

MEMS microaccelerometer Piezoelectric accelerometer

Sensor Type ADXL202JE PCB 393B12

Channels X, Y X

Range (g) 2.0 0.5

Sensitivity (mV/g) 167 17% 10000

Resolution (g rms) 0.002 0.000008

Size (mm) 5.0 x 5.0 x 2.0 30.2 (diam.) x 55.6 (high)

Weight (gram) 1.6 210.0

For DAQ purposes, the Labview software [33] was used to develop a program that records

and transforms the incoming data into engineering units. A subroutine was also made to

pre-process the data by calculating the Welch Spectrum.

Two case studies were carried out aiming at comparing the performance of the

commercial wireless platforms against conventional wired based systems and at assessing

the possibility of their use for OMA of civil engineering structures. Details, results and

comments of the tests are shown next.

4.2 Case study I: Dynamic response of an inverted pendulum

A SDOF structure represented by an inverted pendulum is one of the simplest examples

used by civil engineers to explain the fundamentals of structural dynamics. In this work,

this pendulum is also used to evaluate and understand the behaviour of commercial

wireless-based platforms and their use for civil engineering works.

The studied specimen was an inverted timber pendulum with 1.70 m high and with steel

plates in its top and base (Figure 4a). To perform a complete dynamic characterization of

the pendulum, three wired based and three wireless with embedded MEMS platforms

accelerometers were used (the MEMS sensors were programmed to perform

measurements only in one axis). These sensors were arranged in two setups keeping the

node 1 as common for both measurements, as shown in Figure 4b. For comparison

purposes the wired and wireless systems were set to run concurrently. The DAQ process

was performed at 128 Hz of sampling rate.

(a) (b)

Figure 4 Experimental modal analysis of a laboratory specimen. (a) Studied pendulum

and close up of the sensors arrangement; and (b) setup 01 and 02 description

Initially, the performance with respect to the acceleration time series was studied; using

tests under random excitation and under ambient noise. Figure 5 shows the recorded

signal by mote 3 and accelerometer 3 in both scenarios.

Wireless based

sensors

Wired based

accelerometers

Setup 01 Setup 02

(a) (b)

Figure 5 - Time series collected by mote 3 and accelerometer 3 in the inverted pendulum

tests. (a) Response under random excitation in Setup 01; and (b) response under ambient

noise in Setup 02

The results demonstrate the poor performance of the microaccelerometers for measuring

low amplitude vibrations. The maximum values and the root mean square (RMS)

registered by the wireless platforms are, respectively, 3 to 6 times and 8 to 22 times lower

than the values recorded by the conventional platforms with ambient noise. Similar results

were obtained in the case of the RMS in random excitation, even if the time series

recorded with both systems are rather similar.

Then, the dynamic characteristics of the system were studied. For this purpose, the SSI-

Data method implemented in the ARTeMIS extractor software [34] was used. Figure 6

shows the stabilization diagram corresponding to the random excitation and Table 2

shows a summary of the results accelerometers, where f is the frequency and is the

damping.

(a) (b)

Figure 6 - Stabilization diagrams for the analysis of the time series recorded under random

excitation in the inverted pendulum tests. (a) Results of the conventional wired based

accelerometers; and (b) results of the wireless platforms

Table 2 Results of the experimental modal analysis of the inverted pendulum study.

Conventional

Accelerometers

Wireless

Platforms

Mode (Hz) (%)

(Hz)

(%) fError (%) Error (%)

Random

excitation

1 2.30 1.45 2.35 3.57 2.13 59.39

2 2.71 1.57 2.68 2.94 1.12 46.60

1 2.26 0.82 2.41 9.82 6.22 --

Ambient 2 2.63 2.12 2.83 10.42 7.07 --

According to the frequency content results, the wireless based platforms give accurate

results (errors of about 2% for random excitation and about 7% for ambient vibration).

When the structure is lightly and randomly excited, the modal identification is easier

because the stable poles are properly aligned in the natural frequencies. In the case of

ambient noise the dynamic identification becomes more complicated due to the

appearance of noise poles (stabilization diagrams not shown). The results related to

damping tend to show a large scatter and are often unreliable. Still, no correlation was

found between damping values using conventional and wireless based systems, with

extremely large (and incorrect) values found with the wireless based platforms. Due to

the lack of synchronization algorithms implemented for the motes, no information can be

gathered on the mode shapes.

4.3 Case study II: Dynamic Response of Monuments - The Chimneys at Pao dos

Duques

The Pao dos Duques (Dukes Palace) was built between 1422 and 1433 by D. Afonso

(bastard son of the king of Portugal D. Joao I) in Guimares, north of Portugal. At the

beginning, the building was used as a residence of the deduces of Bragana but then

became mostly unused between 1480 and 1807 [35]. Since 1807, the building was used

as barracks and in 1888 the Architects and Archaeologist Portuguese Society listed it as

a historical monument [36]. In 1937, it was re-built based on available information and

introducing many new elements giving the monument its current impressive character.

Figure 7 shows the present condition and the original condition before the intervention,

in 1937.

(a) (b)

Figure 7 - Pao dos Duques. (a) Present situation [37]; and (b) front view of the palace in

1937 [36]

One of the most important changes in the structure of the building was the addition of

chimneys in the roof. The original building had only four chimneys and, in the

intervention started in 1937, 34 more chimneys were added. Since 2002, the building

suffered some conservation works, mostly related to the roofs and chimneys. The

chimneys exhibited considerable damage, with one chimney requiring strengthening.

Based on the previous results of the experimental tests, the use of commercial wireless

platforms for structural dynamic monitoring was again explored. The dynamic response

of one of the four original chimneys was studied using conventional and wireless

platforms. Figure 8a shows a general view of the conservation works that were carried

out and Figure 8b shows the location of the wireless platforms in the experimental tests.

(a) (b)

Figure 8 - Chimneys at Pao dos Duques. (a) Recent conservation works; and (b) sensors

location

The advantages of using wireless platforms were clear in this case study, as their use is

much simpler. The DAQ process was also easier allowing safer work in a zone with

difficult access.

Figure 9 shows the stabilization diagrams of the analysis of the time series recorded under

random excitation. Table 3 shows the results of the identified frequencies using

conventional accelerometers and wireless platforms.

(a) (b)

Figure 9 - Stabilization diagrams of the analysis of the time series recorded under random

excitation in the chimney at Pao dos Duques tests. (a) Results of the conventional wired

based accelerometers; and (b) results of the wireless platforms.

Table 3 Dynamic response of the chimney at Pao dos Duques

Conv. Accelerometers Wireless Platforms

Mode (Hz) (%) (Hz) (%) Error (%) Error (%)

1 1.69 1.34 1.68 1.61 0.60 16.77

2 1.77 4.22 1.71 0.72 3.51 --

The results show very small differences in the identified frequencies obtained by using

the conventional and the wireless platforms (maximum error is 3.5%). Again,

inconclusive results are obtained with respect to damping. When the tests are performed

with ambient noise (results not shown) similar frequencies could be identified, again with

more difficulties due to the spurious poles appearing in the stabilization diagrams.

5. CONCLUSIONS

This paper explores a new platform, based on wireless technology with embedded MEMS

sensors, for performing operational modal analysis of structures. Commercial WSN

platforms available in the market were chosen for comparison purposes against widely

used conventional wired based systems.

The results of laboratory tests showed that the WSN platforms have poor performance

with respect to the acceleration time series recorded, due to the low resolution of the

microaccelerometers and DAQ systems embedded. The wireless platforms showed also

very poor performance for the detection of modal shapes due to the lack of

synchronization algorithms. In the case of frequency detection, reliable results were

obtained especially when the systems were randomly excited.

In order to study the performance of the wireless platforms in the field, tests were carried

out in the masonry chimneys of a historical 15th century monument in Portugal. Again,

good results were obtained in terms of frequencies identification, with very small

differences found between the frequencies measured with the conventional and the

wireless platforms.

The problems of the commercial wireless platforms and their application for civil

engineering studies have been therefore identified (lack of synchronization and low

resolution). Future developments are needed before the platforms can be used for modal

shape identification and ambient vibration tests of stiff structures.

Acknowledgments

The second author gratefully acknowledges to Alban, European Union Programme of

High Level Scholarships for Latin America, for the financial support with the scholarship

number E07D400374PE.

References

[1]. Ruiz-Sandoval, M.E. (2004), Smart Sensors for Civil Infrastructure Systems. PhD

Dissertation, University of Notre Dame, Notre Dame, Indiana.

[2]. NI (2009), Overview of Structural Health Monitoring Solutions. http://zone.ni.com/,

Accessed on June 2009.

[3]. Arampatzis, T.; Lygeros, J. and Manesis, S. (2005), A Survey of Applications of

Wireless Sensors and Wireless Sensor Networks. Proceedings of the 13th Mediterranean

Conference on Control and Automation.

[4]. Straser, E.G. and Kiremijdian, A.S. (1996), A Modular Visual Approach to Damage

Monitoring for Civil Structures. Proceedings of SPIE Smart Structures and Materials,

2719, 112-122.

[5]. Kiremijdian, A.S.; Straser, E.G.; Meng, T.H.; Law, K. and Soon, H. (1997),

Structural Damage Monitoring for Civil Structures. International Workshop - Structural

Health Monitoring, 371382.

[6]. Straser, E.G. and Kiremijdian, A.S. (1998), A Modular Wireless Damage Monitoring

System for Structures. The John A. Blume Earthquake Engineering Center, 128.

[7]. Straser, E.G.; Kiremijdian, A.S.; Meng, T.H. and Redlefsen, L. (1998), A Modular

Wireless Network Platform for Monitoring Structures. Proceedings of the SPIE

International Society for Optical Engineering, 450-456.

[8]. Lynch, J.P. and Loh, K.J. (2006), A Summary Review of Wireless Sensors and Sensor

Networks for Structural Health Monitoring. The shock and vibration digest, 91-128.

[9]. Crossbow (2009), Crossbow Wireless Sensors Network. www.xbow.com. Accessed

on December.

[10]. Stiharu, I. (2002), MEMS, General Properties. Encyclopedia of Vibration,

Academic Press, 794-805.

[11]. Eloy, J.C. (2007), Status of the MEMS Industry: Evolution of MEMS Markets and

of the Industrial Infrastructure. Sensors & Transducers, 1771-1777.

[12]. Lynch, J.P.; Sundararajan, A.; Law, K.H.; Carryer, E.; Farrar, C.R.; Sohn, H.; Allen,

D.; Nadler, B. and Wait, J.R. (2004), Design and Performance Validation of a Wireless

Sensing Unit for Structural Health Monitoring Applications. Structural Engineering and

Mechanics, 17 (3), 393-408.

[13]. Lynch, J.P. (2007), An Overview of Wireless Structural Health Monitoring for Civil

Structures. Philosophical Transactions of the Royal Society of London Series A,

Mathematical and Physical Sciences, 345-372.

[14]. Wang, Y. (2007), Wireless Sensing and Decentralized Control for Civil Structures:

Theory and Implementation, PhD Dissertation, Stanford University, USA.

[15]. Severino, R. (2008), On the use of IEEE 802.15.4/ZigBee for Time - Sensitive

Wireless Sensor Network Applications. Master Dissertation, Polytechnic Institute of

Porto, Porto.

[16]. TinyOS (2009), www.tinyos.net, Accessed on March 2009.

[17]. Olusola, F. and Larsson, T. (2007), The Evaluation of TinyOS with Wireless Sensor

Node Operating System. Halmstad University, Sweden.

[18]. Gay, D.; Levis, P.; Von Behren, R.; Welsh, M.; Brewer, E. and Culler, D. (2003),

The nesC Language: A Holistic Approach to Networked Embedded Systems. PLDI '03,

San Diego, California.

[19]. Aoki, S.; Fujino, Y. and Abe, M. (2003), Intelligent Bridge Maintenance System

Using MEMS and Network Technology. Smart Systems and NDE for Civil Infrastructures

- SPIE, San Diego, CA, 5057, 37-42.

[20]. Lynch, J.P.; Sundararajan, A.; Law, K.H.; Kiremijdian, A.S.; Carryer, E.; Sohn, H.

and Farrar, C.R. (2003), Field Validation of a Wireless Structural Health Monitoring

System on the Alamosa Canyon Bridge. Smart Systems and NDE for Civil Infrastructures

- SPIE, San Diego, CA, 5057 267-278.

[21]. Chung, H.C.; Enomoto, T.; Shinozouka, M.; Chou, P.; Park, C.; Yokoi, L. and

Morishita, S. (2004), Real-time Visualization of Structural Response with Wireless

MEMS Sensors. 13th World Conference on Earthquake Engineering, Canada.

[22]. Lynch, J.P.; Wang, Y.; Law, K.H.; Yi, J.-H.; Lee, C.G. and Yun, C.B. (2205),

Validation of a Large-scale Wireless Structural Monitoring System on the Geumdang

Bridge. Proceedings of the International Conference on Safety and Structural Reliability

ICOSSAR, Rome, Italy.

[23]. Weng, J.-H.; Loh, C.-H.; Lynch, J.; Lu, K.-C.; Lin, P.-Y. and Wang, Y. (2008),

Output-Only Modal Identification of a Cable-Stayed Bridge using Wireless Monitoring

Systems. Engineering Structures, 30, 1820-1830.

[24]. Pakzad, S.N.; Fenves, G.L.; Kim, S. and Culler, D. (2008), Design and

Implementation of Scalable Wireless Sensor Network for Structural Monitoring. Journal

of Infrastructure Systems, 89-101.

[25]. Ou, J.P.; Li, H.; Xiao, Y.Q. and Li, Q.S. (2005), Health Dynamic Measurement of

Tall Building Using Wireless Sensor Network. Proceedings of the Smart Structures and

Materials-SPIE, San Diego, CA, USA, 205-215.

[26]. Ceriotti, M.; Mottola, L.; Picco, G.P.; Murphy, A.L.; Guna, S.; Corra, M.; Pozzi,

M.; Zonta, D. and Zanon, P. (2009), Monitoring Heritage Buildings with Wireless Sensor

Networks: The Torre Aquila Deployment. International Conference on Information

Processing in Sensor Networks-IPSN09, San Francisco, USA.

[27]. Cunha, A.; Caetano, E.; Magalhes, F. and Moutinho, C. (2006), From Input-Output

to Output-Only Modal Identification of Civil Engineering Structures, SAMCO.

[28]. Rodrigues, J. (2004), Identificao Modal Estocstica Mtodos de Anlise e

Aplicaes em Estruturas de Engenharia Civil. PhD Dissertation, FEUP/LNEC, Lisbon,

Portugal.

[29]. Van Overschee, P. and De Moor, B. (1991), Subspace Algorithms for the Stochastic

Identification Problem. Proceedings of the 30th Conference on Decision and Control,

Brighton, England.

[30]. Peeters, B. and De Roeck, G. (1999), Reference-Based Stochastic Subspace

Identification for Output-Only Modal Analysis. Mechanical Systems and Signal

Processing, Academic Press, 855-878.

[31]. PCB (2009), Product Catalog. www.pcb.com, Accessed on December 2009.

[32]. NI (2009), User Guide and Specifications. www.ni.com, Accessed on December

2009.

[33]. Labview (2006), Labview User Manual, Release 8.0. National Instruments

Corporation, USA.

[34]. SVS (2006), ARTEMIS extractor pro user manual, Release 3.5. Structural Vibration

Solutions.

[35]. Azevedo, A. (1964), Pao dos Duques. Livraria Lemos, Guimares, Portugal.

[36]. Silva, J.H.P. (1974), Pao dos Duques em Guimares.

[37]. Bing-Maps (2009), http://www.bing.com/maps/. Accessed on December.