POST-ACCIDENT DAMAGE IDENTIFICATION BASED...

POST-ACCIDENT DAMAGE IDENTIFICATION BASED ON VIBRATION MEASUREMENTS APPLIED TO MASONRY STRUCTURES

Luís F. Ramos (1) and Paulo B. Lourenço (2)

(1) ISISE, University of Minho, Portugal, [email protected]

(2) ISISE, University of Minho, Portugal, [email protected]

AbstractThe present paper aims at damage assessment of masonry structures in an early stage.

Two replicates of historical constructions were built in virgin state, one arch with 1.5 m

span and one shear wall of 1 m2. Afterwards, progressive damage was applied and sequential

modal identification analysis was performed in each damage stage, aiming at finding adequate

relations between changes in dynamical behaviour and internal crack growth.

During the dynamic tests, accelerations and strains were recorded in many points of the

replicates. Comparisons between different techniques based on vibrations measurements were

made to evaluate which methods are the most suitable for identifying damage in masonry

constructions.

All the knowledge emerged from the experimental tests was then applied to two case

studies: a masonry clock tower, in the northern part of Portugal, and the church of Monastery

of Jerónimos, in Lisbon. The two monuments have installed a dynamic monitoring system,

where the environmental effects are measured.

The paper presents the results of the arch model simulation and the results of the case study

of Jerónimos Monastery church.

Keywords Damage identification, modal testing, masonry

1. INTRODUCTION

The present paper deals with the problem of damage identification by using Global and

Local damage identification techniques. It is advantageous to have two categories of damage

453

on Site Assessment of Concrete, Masonry and Timber Structures - SACoMaTiS 2008

1-2 September 2008, Varenna, Italy

assessment methods: (a) the vibration based damage identification methods, currently defined

as Global methods, because they do not give sufficiently accurate information about the

extent of the damage, but they can identify its presence and define its precise location [1];

and (b) the methods based on visual inspections or experimental tests, such as sonic or

ultrasonic methods or radar methods [2], also called as Local methods. The latter need a

preceding global approach (Global methods) to detect and localize the damage, and then,

if the possible location of damage is accessible in the structure, they can describe the damage

in an accurate way.

Damage on masonry structures mainly relates to cracks, foundation settlements, material

degradation and displacements. When cracks occur, generally they are localized, splitting the

structures in macro-blocks. Dynamic based methods to assess the damage are an attractive

tool for this type of structure due to the present requirements of unobtrusiveness, minimum

physical intervention and respect of the original construction. The assumption that damage

can be linked to a decrease of stiffness seems to be reasonable for this type of structure.

Many methods are presented in the literature, see [3], for damage identification based on

vibration signatures but there are only a few papers on the application to masonry structures.

An important task before damage can be identified from vibration characteristics is the study

and subsequent elimination of the environmental effects [4], which for masonry structures can

have significant importance [5].

1. PROPOSED METHODOLOGY

During the last decade, software and hardware developments made continuous monitoring

possible nowadays [6]. Typically, one can install hundreds of sensors in a structure and read

the data in real time. The attention now is focused on the following aspects: what type of

information is important from the structural point of view and how the data should be

processed and stored for damage analysis [7]. The developments in MEMS and WSN are

promising technologies in this field, in particular for historical masonry structures.

Concerning monitoring of historical masonry structures, this task can be divided in four

phases:

1. The first phase is the data collection of the structure, including the historic

information, geometrical and topographic survey, damage survey, the mechanical

materials characterization with ND tests, a global dynamic modal test and a

numerical model analysis for static and dynamic calibration. This is the first

approach to the structural behaviour in the assumed healthy condition at time

“zero”;

2. In the second phase the health monitoring plan can be performed with a limited

number of sensors (e.g. a pair of reference accelerometers, strain gauges at critical

sections, temperature and humidity sensors, etc). Data should be stored periodically

and the monitoring system should be able to send the proper alarms. Environmental

effects should be studied and the presence of damage should be observed by the

global modal parameters;

3. In the third phase, after alarm triggering or an accident, a full-scale dynamic survey

with more sensors and measuring points should be performed. In this phase the

“health condition” of a structure is studied with more detail. Damage identification

methods should be applied to the structure after filtering the environmental effects.

454



The aim of the dynamic methods is to confirm and locate the (possible) damage in a

global way; and

4. In the last phase, a local approach with visual and complementary ND tests should

be performed to locally assess the damage and classify it. This can be carried out

with sonic test or radar tests, depending on the access conditions of the structure.

This local approach can give a better definition of the damage.

For the third phase, a group of damage methods has been selected from the literature.

In one hand, it is intended to study the applicability of existing methods to the masonry

structures, and, in another hand, it is aimed to have a wide view of the problem (different

results are provided by different methods), assisting in the conclusions related to damage

identification. If significant damage is present in the structure, the results provided from

different methods would converge in a unique conclusion, giving more confidence to the

analyzer. The selected methods together with the required modal information are presented in

Table 1 (see [3] for the complete description of each method).

Table 1: Selected damage identification methods

Modal Information Method Type

Expected Identif.

Level

Comparison to a

Ref. Scenario ω ϕ ϕ″ φ φ″Unified Significance

Indicator (USI) Level 1 Yes •

COMAC Level 2 Yes

Parameter Method (PM) Level 2 Yes •Mode Shape Curvature

Method (MSCM) Level 2 Yes

Damage Index Method

(DIM) Level 2 Yes

Sum of the Curvature

Errors method (SCE) Level 2 Yes

Change Flexibility

Matrix method (CFM)

Non-Model Based

Level 2 and 3 Yes • •

FE Model Updating

method (FEMU) Model Based

Level 2 and 3 No

– Optional modal quantities; • – Compulsory modal quantities;

Level 1 – Detection; Level 2 – Localization; Level 3 – Assessment

All methods have one common aspect; they all use spatial modal information of the

structure, through the mass scaled or non-scaled mode shapes φ and ϕ, respectively

(or/and through the mass scaled or non-scaled curvatures mode shapes φ″ and ϕ″,

respectively).

The damage methods were applied to the experimental models, where progressive and

controlled damage scenarios were imposed. From the point of view of the applicability of

dynamic based identification methods to masonry structures, the methodology would be

successful if the detection (Level 1), the localization (Level 2) and the assessment (Level 3)

will be attained with these methods. The monitoring approach was applied to two real case

studies: the Mogadouro Clock Tower, in Portugal, and the Church of the Monastery of

Jerónimos, in Lisbon.

455



3. LABORATORY SIMULATIONS

One replicate of ancient masonry arches was built in the laboratory with clay bricks with

100 × 50 × 25 mm3. The bricks were handmade in the Northern area of Portugal. The clay

brick, with low compression strength, and a mortar with poor mechanical properties used for

the joints tries to be representative of the materials used in the historical constructions.

The arch has a semicircular shape with a radius of 0.77 m, a span of 1.50 m, a width of

0.45 m, and a thickness of 0.05 m, and rests in two concrete abutments fixed to the ground

floor with bolts. All the tests were carried out after 60 days of the specimen construction.

3.1 Static Tests Progressive and controlled damage were applied by static increasing loads to reach

multiple damage levels (several cracks). The loads were applied and removed with linear

branches for all the levels. Between each stage (damage scenario), modal identification

analysis using out-put-only (ambient or natural vibration) techniques were performed [4],

where the ambient temperature and humidity were also recorded, to evaluate possible

environmental effects on the dynamic response.

In the arch eight Damage Scenarios (from DSI to DSVIII) were induced. Figure 1a shows

the load application point and the resulted four cracks (c1, c2, c3 and c4). Figure 1b shows the

response of the model during the subsequent static tests, where it is possible to visualize the

probable occurrence of the cracks and the stiffness decrease after each damage scenario.

Figure 2c presents one of four cracks found in the arch. It should be stressed that the

maximum remaining crack opening after the applied loads was 0.05 mm and the maximum

crack depth in the loading branch was 30 mm for crack c1 (more than half of the arch

thickness).

(a)

Arch Static Response

0.0

0.5

1.0

1.5

2.0

0.0 0.2 0.4 0.6 0.8 1.0Vertical Displacement [mm]

Ver

tica

l L

oad

[k

N]

c 1c 2 c 3

c 4

k 0 k 1 k 2

k 3

k 4

(b) (c)

Figure 1: Arch static tests: (a) crack locations and sensor positions; (b) static structural

response; and (c) crack c2 in the intrados

3.2 Dynamic Tests In order to have a clear definition of the modal displacements, it was decided to make the

measurements in 11 points uniformly distributed along the arch. The 11 points were

materialized along two lines at the specimen sides for the accelerometers and along the

specimen centre line for the strain gauges. In total, 44 different directions for accelerations

(each side, in radial and tangential directions) and 22 strain points (intrados and extrados, in

tangential direction) were measured. Figure 2 shows some images of the sensors location in

the arch.

456



(a) (b) (c)

Figure 2: Arch dynamic tests: (a) sensors location; (b) strain gauges;

and (c) normal and tangential accelerations measurements

The accelerometers were bolted to aluminum plates that were directly glued to the

specimens. The strain gauges have 12 cm length. The dynamic tests were performed under

two different type of excitations: ambient and random impacts introduced by an impact

hammer. Only with random impact excitation it was possible to measure with accuracy the

modal strains in the models.

3.3 Damage Analysis The damage identification analysis was divided in three parts: the analysis of the global

parameters changes, the analysis with non-model based methods and the analysis with the

Finite Element model updating method. Here, only the first two analyses will be presented.

Table 2 presents the frequency results for the progressive damage scenarios and Figure 3a

gives the relative variation of the frequencies. Observing the global frequency results,

the modal properties of the masonry specimens seem sensitive to the damage progress.

The residual values in the last scenario are between 78 and 95% of the reference values.

These results seem promising, as other tests in the literature report smaller changes in

frequencies values [4]. Another global parameter to study is the damping coefficient. Here,

a significant increase of damping was observed after DSIV, see Figure 6b, where the average

values for the damping coefficients using 6 and 7 mode shapes are presented.

Table 2: Frequency results for the arch model with ambient excitation

Mode 1 Mode 2 Mode 3 Mode 4

ω CV ∆ω ω CV ∆ω ω CV ∆ω ω CV ∆ωDamage Scenario

[Hz] [%] [Hz] [Hz] [%] [Hz] [Hz] [%] [Hz] [Hz] [%] [Hz]

RS 35.59 0.57 – 67.30 0.69 – 72.11 0.53 – 125.74 0.52 –

DSI 35.55 0.44 –0.05 67.51 0.61 +0.21 71.80 0.27 –0.30 125.69 0.76 –0.05

DSII 35.55 0.34 –0.04 67.39 0.83 +0.09 71.83 0.74 –0.28 125.79 0.81 +0.05

DSIII 35.42 0.44 –0.17 67.47 0.88 +0.17 71.66 0.66 –0.45 125.75 0.88 +0.01

DSIV 35.15 0.34 –0.44 67.11 0.66 –0.19 71.33 0.41 –0.78 126.01 0.43 +0.28

DSV 33.72 0.48 –1.87 65.68 0.54 –1.62 69.36 0.43 –2.75 124.48 0.64 –1.25

DSVI 33.19 0.52 –2.40 64.91 0.79 –2.39 68.56 0.42 –3.55 123.58 0.56 –2.16

DSVII 31.49 0.69 –4.10 63.08 1.02 –4.22 65.72 0.52 –6.39 121.97 0.75 –3.77

DSVIII 28.09 1.11 –7.50 58.44 1.20 –8.86 62.61 0.74 –9.50 119.44 0.74 –6.30

– Damage scenario where first visual the crack (c1) were localized

457



Relative Frequency (Ambient)

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

RS

DS

I

DS

II

DS

III

DS

IV

DS

V

DS

VI

DS

VII

DS

VII

I

Damage Scenario

Rel

ativ

e F

req

uen

cy

Mode 1 Mode 2 Mode 3Mode 4 Mode 5 Mode 6Mode 7

(a)

Relative Damping (Ambient)

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

RS

DS

I

DS

II

DS

III

DS

IV

DS

V

DS

VI

DS

VII

DS

VII

I

Damage Scenario

Rel

ativ

e D

amp

ing Average with 7 Modes

Average with 6 Modes

(b)

Figure 3: Dynamic global response of the arch model: (a) relative frequency variation;

and (b) relative damping variation

Concerning the other non-model based methods (see Table 1) the seeking of the damage

location was based in the analysis of the results from the methods which gave consistent

results, namely the MSCM, the DIM and the SCE.

Figure 4: Damage location for the arch model: (a) comparison with the RS; (b) consecutive

comparison with each DS; and (c) comparison with DSIII as a new RS

RS – DSI RS – DSI

Legend:

Crack detected in the

current comparison

RS – DSII DSI – DSII

Crack detected in the

preceding comparisons

Observed cracks

RS – DSIII DSII – DSIII

P3

RS – DSIV

P3

DSIII – DSIV

P1

P7

P3DSIII – DSIV

P7

P1

RS – DSV

P3

P1DSIV – DSV

P1

P7

P3DSIII – DSV

P7

P1

RS – DSVI

P3

P1DSV – DSVI

P3

P7

P1

P9

P1

DSIII – DSVI

P7

P1

RS – DSVII

P3

P1 P11DSVI – DSVII

P3

P7

P1

P9

DSIII – DSVII

P7

P1

P7

P11

RS – DSVIII

P3

P1

P9

P11DSVII – DSVIII

P3

P7

P1

P9

P11DSIII – DSVIII

P7

P1

P9

P11

(a) (b) (c)

458



Figure 4 presents the final location results for three different comparisons. From Figure 4 it

was possible to conclude that the cracks could be localized (level 2) in the vicinity of the

experimental cracks. Considering the above results, it seems that the combination of several

damage methods based on experimental modal curvatures is a good methodology to detect

and locate accurately and at an earlier stage the damage in the case of the masonry arch.

4. CASE STUDY: CHURCH OF MONASTERY OF JERÓNIMOS

The Monastery of Jerónimos, located in Lisbon, is one of the most famous Portuguese

monuments (see Figure 5a). The authentic Portuguese “Manuelino” architectural style of the

XVI Century makes the monument very attractive for tourists. The church of the monastery,

Santa Maria de Belém church, has considerable dimensions: a length of 70 m, a width of 40 m

and a height of 24 m (see Figure 5b).

(a)

0.0 25.0 m

A A

B

B

C

C

(b)

Figure 5: Damage location for the arch model: (a) comparison with the RS; (b) consecutive

comparison with each DS; and (c) comparison with DSIII as a new RS

4.1 Modal Identification Twenty points of the main were selected to measure the modal response, see Figure 6a.

Two output-only techniques were applied to compare the experimental dynamic:

the Enhanced Frequency Domain Decomposition (EFDD) and the Stochastic Subspace

Identification (SSI) method, see [4] and [5].

P2 P3 P4 P5 P6

P29 P15 P14 P13 P12 P28

P21 P20 P19

P30 P22 P24 P25 P26 P27

P10 P9 P8

P16 P17 P18

P23

P7P11

P1

(a) (b)

Figure 6: Dynamic test: (a) location of the measuring points; (b) sensor at the nave extrados,

under the roof tiles

xy

459



Table 2 summarizes eight natural frequencies, damping ratios and MAC values estimated

by the two output-only techniques. The natural frequencies range from 3.7 to 15.1 Hz and no

significant differences could be found between the two methods. For the damping

coefficients, differences up to 140% can be observed. The Modal Assurance Criteria (MAC)

values are only higher than 0.95 for the first two modes as a consequence of the difficulty in

exciting this heavy structure.

Table 3: Experimental modal results

ω[Hz]

ξ[%]

Mode

Shape EFDD SSI EFDD SSI

MAC

1st 3.69 3.68 2.34 1.26 0.99

2nd 5.12 5.04 1.11 2.68 0.92

3rd 6.29 6.30 1.00 0.82 0.67

4th 7.23 7.29 0.77 1.44 0.67

5th 9.67 9.65 1.10 1.45 0.62

6th 11.64 11.65 1.20 1.46 0.36

7th 12.45 12.51 1.25 1.19 0.71

8th 14.99 15.09 1.31 2.77 0.49

Figure 7a shows the average normalized single value of all data calculated with the EFDD

method, were is possible to identify clearly the peaks of the resonant frequencies. Figure 7b

presents the mode shape configuration for the first resonant frequency. The first mode is

manly a transversal mode of the main nave, with significant vertical amplitudes in the central

part, as a result of effect of the slenderness columns.

Frequency [Hz]

0 5 10 15 20 25-40

-30

-20

-10

0

10

20

(a) (b)

Figure 7: Modal estimation: (a) normalized single values of the EFDD method;

(b) first mode shape

4.2 Monitoring Task A dynamic system is in operation since April, 2005. The system is composed by two

strong motions recorders with 16 bits ADC analyzers provided with batteries. One triaxial

force balance accelerometer is connected to each recorder by cable. The accelerometers have

a bandwidth form DC to 100 Hz, a dynamic range µ1 g, a sensitivity of 10 V/g and an

operating temperature range from 20 to 70ºC. The two devices connected (sensor and

analyzer) give a final resolution of 8 µg.

460



Two points were selected to install the sensors (see Figure 8): one sensor (A1) was

installed on the base of the structure near the chancel, and the other sensor (A2) on the top of

the nave and in the extrados, in the same location of point P1 of the modal identification

analysis of the nave (see Figure 6). The sampling frequency is equal to 100 Hz.

0.0 25.0 m

A2A1, R

(a) (b)

Figure 8: Monitoring system: (a) sensors and recorders location; and (b) sensor (A1) and

recorders (R)

The monitoring system was design to record large and also micro-earthquakes.

The available data at the moment is based on triggered events of 1.5 min long and on a set of

non-periodic events of 10 min during some selected days.

4.3 Damage Detection and Environmental Effects As no continuum series were recorded in the monitoring system, modelling of thermal

inertia is rather difficult and only the static regression models were used. From the observed

bilinear trend, two linear regressions for temperature values lower and higher than 17.5ºC

were adopted. Figure 9 shows the results for the first estimated frequency by correlating the

two quantities, temperature and frequency, and by plotting in time the evolution of the

frequencies and the model with the 95% (±2σ) confidence intervals.

Static Regression for the First Frequency

3.5

3.6

3.7

3.8

3.9

4.0

4.1

4.2

0 5 10 15 20 25 30 35 40

Temperature [ºC]

Fre

quen

cy [

Hz]

(a)

Static Model for the First Frequency

3.5

3.6

3.7

3.8

3.9

4.0

4.1

4.2

Au

g-0

5

Oct

-05

Dec

-05

Feb

-06

Apr-

06

Jun-0

6

Au

g-0

6

Oct

-06

Dec

-06

Feb

-07

Apr-

07

Jun-0

7

Au

g-0

7

Oct

-07

Date

Fre

qu

ency

[H

z]

(b)

Figure 9: Environmental effects: (a) bilinear relation between temperature and the first

frequency; and (b) frequency shifts among time

461



The results show that the bilinear static model follows the evolution of the frequencies but

a significant number of outliers can be observed in Figure 9. This indicates that others

environmental effects need to be studied to better model the dynamic response.

On 12 February, 2007, at 10:35 am, a 5.8 magnitude earthquake occurred in the Southwest

of Lisbon. The permanent staff of the monument felt the ground shake. No visitors were

inside the church because on Mondays the church is closed to public. The strong motion

recorder at the base (A1) measured a PGA of 2 mg, but the estimated natural frequencies did

not suffer any significant shift, as can be observed for the case of the first one, presented in

Figure 9. Therefore, no damage in the structure occurred due to this minor earthquake.

5. CONCLUSIONS

In the paper it was proposed a methodology to identify damage after and significant event.

This methodology was applied to on arch model in the laboratory and to a real case study, the

church of Jerónimos Monastery, in Lisbon.

During the several damage scenarios in the arch model, induced by controlled loads, it was

possible to observe that natural frequencies were sensitive to progressive damage. Although

the damage in the specimens was very difficult to visualize, the static response exhibits an

evident decrease of stiffness and the natural frequency values decrease significantly.

The problem of the damage detection was solved only with the frequency shifts observation.

The dynamic monitoring system installed in the monastery of Jerónimos allowed observing

that temperature has a significant effect on the dynamic response, while the excitation level

could be neglected. The temperate and frequencies exhibit a bilinear correlation, with

frequencies varying almost 6%, on average.

From the experience with the case, the proposed methodology for damage identification

seems to be useful and applicable to masonry structures, especially to historical constructions.

The frequency observation seems to be a reliable quantity for damage detection.

REFERENCES

[1] Chang, P.C.; Flatau, A.; Liu, S.C., ‘Review Paper: Health Monitoring of Civil Infrastruc-ture’,

Structural Health Monitoring 2 (3) 257-267, 2003.

[2] Doherty, J.E., ‘Nondestructive Evaluation, Handbook on Experimental Mechanics’, A.S.

Kobavashi Edt., Society for Experimental Mechanics, 1987.

[3] Doebling, S.W.; Farrar, C.R.; Prime, M.B.; Shevitz, D., ‘Damage identification and health

monitoring of structural and mechanical systems from changes in their vibration characteristics: a

literature review’, Los Alamos National Laboratory, NM, 1996.

[4] Peeters, B., ‘System Identification and Damage Detection in Civil Engineering’, PhD Thesis,

Catholic University of Leuven, Belgium, 2000.

[5] Ramos, L.F., ‘Damage Identification on Masonry Structures Based on Vibration Signatures’,

PhD Thesis, University of Minho, www.civil.uminho.pt/masonry, 2007.

[6] Maeck, J., ‘Damage Assessment of Civil Engineering Structures by Vibration Monitoring’,

PhD Thesis, Catholic University of Leuven, Belgium, 2003.

[7] Londoño, N.A., ‘Use of Vibration Data for Structural Health Monitoring of Bridges’, PhD Thesis,

Carleton University, Ottawa, Canada, 2006.

462

Date post:	05-Jun-2018
Category:	Documents
Upload:	phungquynh
View:	212 times
Download:	0 times

POST-ACCIDENT DAMAGE IDENTIFICATION BASED...

Documents