Structural System Identification From Ambient and Forced Vibration Testing
Structural System Identification From Ambient and Forced Vibration Testing
Structural System Identification From Ambient and Forced Vibration Testing
N. Haritos
E-mail: nharitos@unimelb.edu.au
Telephone: + 61-03-83446829; Fax: +61 -03-83444616
Abstract: For over two decades now the author and his team have accumulated wide-ranging experience in the
dynamic testing of structural systems (such as flooring systems in buildings and stadia, pedestrian and road
bridges and other structural forms such as gantry frames) for the purposes of their condition and performance
assessment. Identification of the in-service state of a structure is important:
• for determining its basic articulation and the nature of the conditions at its supports;
• for determining its structural performance characteristics whilst in service;
• as a precursor to performing an assessment of its load carrying capacity;
• for gauging the effect of a retrofit that may have been introduced on it;
• for performing general condition monitoring and assessment to gauge the effects of any degradation
over a period of time or following a specific possibly damaging event, (eg an earthquake, storm or accidental
load). This paper considers several examples of application of dynamic testing drawn from this experience to
illustrate the utility of the approaches adopted concentrating on (but not restricted to) bridge engineering
applications
Keywords: dynamic testing; experimental modal analysis; vibration modes; structural health monitoring
1. Introduction
Vibration-based assessment techniques for estimating the structural “health” or in-situ condition or
damage of bridges and other structures has been receiving increasing attention in the engineering
scientific community in recent years, [1, 2, 3]. The basis behind the techniques adopted for doing this
is the simultaneous recording of vibratory response, resultant from some sort of forcing stimulus,
taken over a pre-designed grid of measurement points on the structure. When the forcing stimulus is
able to also be contemporaneously measured and recorded with the vibratory response, then it is
possible to perform traditional frequency-domain based Experimental Modal Analysis (EMA) on the
data capture, [4], whereas if this is not possible, then alternative time-domain based methods or
approximate frequency-domain based methods can be used, [5, 6] for performing the EMA. Ambient
vibration is the term coined for where the excitation stems from the normal operating conditions of
the structure of interest eg, wind action on a tall building; wave action on an offshore structure; traffic
excitation on a road bridge and/or ground motion induced by traffic on roadways underneath a flyover
bridge; pedestrian excitation of a footbridge or floor system, etc. Forced vibration is the term coined
for where the excitation is purpose introduced through a shaker system (eg electromagnetic/ hydraulic
shaker) or an impact hammer/device. In the case of forced excitation the input forcing function and its
characteristics are user controlled/specified and able to be measured whereas for ambient excitation, it
is not normally possible to measure the excitation forcing function and its characteristics.
The primary function of EMA is to essentially produce a set of mode shapes and their associated
natural frequencies of vibration and damping levels from the original contemporaneously measured
data capture after suitable conditioning and transformation. Information gleaned from the modes of
vibration themselves or from the transformed original data can then be used to perform a “health
assessment” of the structural system to which the data capture corresponds. This could be in the form
of a direct comparison of the modes themselves with those predicted by a Finite Element Analysis
(FEA) model in which differences between these can be used to perform FEA model updating of
modelling parameters. Alternatively, a number of approaches that operate on the measured response
characteristics to determine a Damage Index can also be exercised in an attempt to extend the
condition assessment to the point of identifying the damage location(s) and degree of severity.
{ } given by:
shape ψ j
{φi } {ψ j }
T 2
MACij = (1)
{φi } {φi }{ψ j } {ψ j }
T T
MAC values greater than 0.9 reflect a high degree of correlation (with 1.0 being a “perfect” fit) and
values less than 0.1 associated with uncorrelated (virtually orthogonally disposed) modes.
Model Updating
Parameters No
Dynamic
Test
iii. A Data Acquisition System (DAS) with anti-aliasing filters, simultaneous sample and hold,
modules for transducer signal-conditioning and excitation, and data storage/processing features
In addition to the above hardware set, suitable software to analyse the data capture for the key
dynamic characteristics of the system under test, would also be necessary. The degree of
sophistication of the hardware used for the data capture and the algorithms adopted in the analysis
software is dependent upon the nature of the structural system under investigation and the degree to
which the information content in the data capture is to be explored by the analyst.
In essence, the choice of analysis technique is dependent upon the particular conditions at hand. The
experience gained with performing structural system identification via vibration response
measurements of bridge, beam/frame and floor systems, and even trees, by The University of
Melbourne, has embraced the full range of possibilities outlined above, [17, 18]. Some examples
drawn from this experience, featuring key results, are presented in the sections that follow.
2.1 Application to typical simply supported span of multi-span RC deck on steel girder road bridge
A typical nominally simply supported span of McCoy’s Bridge over the Goulburn River was
dynamically tested so as to identify its in-service condition and verify the integrity of the composite
action of the RSJ girders imbedded in the RC deck slab. Figure 2 depicts some of the features of this
field test experience with a view of the hydraulic shaker mounted on the bridge deck, the 16 second
long traces of vibration measurement taken contemporaneously with the force trace (a type of Swept
Sine Wave forcing) and the associated ensemble averaged FRF function for this accelerometer (from
16 repeat test records) typical of an internal point from the 7 x 7 measurement grid adopted. Figure 3
depicts a photo of the bridge with its multiple simply supported spans, and the first three modes
identified from DSMA which compare favourably with FEA predictions for a “healthy” bridge deck.
Structural system identification here has verified the integrity of the composite action between the
steel girders and the bridge deck, despite the age of the bridge at the time, being over 60 years.
2.2 Application to typical continuous span slab on beam bridge over Concongella Creek
Figure 4(a) depicts the Concongella Creek Bridge near Stawell, Victoria. This RC deck-on-beam
bridge is over 60 years old and consists of three continuous spans and is meant to have been
constructed en-castre with its abutment ends. Dynamic testing using the LHS and EMA via DSMA
software enabled identification of several modes of vibration, the first three of which are depicted in
Fig 4(b). It is noted that DSMA was capable of distinguishing Modes#2 and #3 despite these being so
close in frequency. In addition, it is also clear from the observed modes that the fixity at the abutment
ends has deteriorated to now act virtually as pins. The enhanced torsional stiffness due to aggregate
interlock is reflected as a slightly higher modal frequency in Mode#2 than is predicted by FEA,
whereas the observed and predicted modal frequencies of the flexural modes are in closer agreement.
(a) 2 (m/s )
2 (b)
"M"
-2
(kN)
60
Force
-60
0 5 10 15
Time (sec)
(c)
Amplitude (m/N)
0.21m
(d)
Phase (deg.)
4 @ 1.3 = 5.2m
2.6m
7.6m
shake
2.6m
RSJs
Frequency (Hz)
6 @ 2.53 = 15.2m
Figure 2: (a) Hydraulic Shaker unit (b) Typical accelerometer and Forcing traces (c) FRF details for
a selected accelerometer on a test span of McCoy’s Bridge (d) Grid of accelerometer locations
(a) (b)
Figure 3: (a) View of McCoy’s Bridge (b) Comparison of EMA and FEA modal predictions
(a) (b)
Figure 4: (a) View of Concongella Creek Bridge (b) Comparison of EMA and FEA modal predictions
Dynamic response measurements were performed over a grid of 16 measurement points on one
cantilever span before remedial work took place and repeated on the opposite balanced span when the
bridge was part open with cantilever ends free, (first test series). These measurements were repeated
at a later date, post installation of the refurbished timber deck, for the part open and fully closed
conditions, (second test series). Model updating of an FEA model suggested that the effective soil
stiffness determining the stiffness of the central pier and influencing the modal characteristics of the
bridge as a whole was virtually mid range to the values inferred from soil tests at these piers, [19]. In
addition, the effect of replacing the timber deck essentially improved the torsional stiffness of the
cantilever sections leading to a higher torsional mode frequency compared to the original deck where
timber planks were rather loosely fitting on the deck. Figure 5(b) provides insight into the modal
results of the tuned/updated FEA model by providing a comparison with those observed from SEMA.
(a) (b)
Figure 5: (a) Sale Swing Bridge (b) Comparison of SEMA results pre and post restoration works
(b)
(a)
F G H
5 6 7
A B C D E I
0 1 2 3 4 8
J
9
0.024
Sa(f) (m2/s3) (c) (d)
0.020
0.016
0.012
0.008
0.004 f (Hz) fo = 4.88, 5.25,
0.000
0 10 20
5.51 Hz
Figure 7(a) and Figure 7(b) depict a photograph of a typical gantry frame and the acceleration
response spectra for the X (longitudinal), Y (vertically transverse) and Z (horizontally transverse)
directions for Gantry frame “G3”, respectively. The primary mode frequencies in the three mutually
orthogonal response directions X, Y and Z directions of 1.18 Hz, 1.47 Hz and 1.24 Hz are clearly
distinguished as rather sharp peaks, reflecting the very low associated damping values of 0.4%, 0.3%
and 0.8% critical, respectively. These conditions would suggest that a mean wind speed of 13.6 m/s
incident at right angles to the plane of gantry frame G3 would correspond to a Strouhal number of
0.13 that would be associated with the possibility of vortex shedding with the potential of causing
resonant vertical vibration on this gantry frame. The signage and other local attachments on the frame
would likely disrupt the formation of any regular vortex street, but this contention remains to be
further investigated.
4. Concluding remarks
This paper has considered a range of dynamic testing methods of varying degrees of sophistication
aimed at performing some sort of structural system identification. Model updating techniques for
tuning FEA model parameters to obtain high correlation of predicted modal characteristics to those
observed from EMA or Simplified EMA (SEMA) have been overviewed and examples presented
drawn from the author’s experience of application to road bridges and the MCG grandstand. The
versatility of newly developed compact tri-axial accelerometers with on board data-logging
capabilities has been exemplified in this paper through reporting of their recent successful use for
investigating the primary mode characteristics of slender gantry frames over the Monash freeway.
Figure 7: (a) View of typical gantry frame (b) Ambient response spectra (X, Y, Z directions)
References
1. Mottershead, J.E., Friswell, M.I., “Model updating in structural dynamics: A survey”. Journal of Sound and
Vibration, 1993, 167:347-375.
2. Farrar, C. R., Doebling, S. W. & Nix, D. A., “Vibration-based structural damage identification”, Phil.
Trans. R. Soc. Lond. A, 2001, 359, 131-149.
3. Sohn, H., Farrar, C. R., Hemez, F. M., Shunk, D. D., Stinemates, D. W., Nadler, B. R., & Czarnecki, J. J.,
“A review of structural health monitoring literature: 1996-2001”, Los Alamos National Laboratory Report,
LA-13976-MS, 2004.
4. Ewins, D. J. Modal Testing: Theory and Practice, New York: John Wiley, 1985.
5. Haritos, N. & Abu-Aisheh, E. “Dynamic testing techniques for structural identification of bridges”, Proc.
ASEC'98, Auckland, 1998, Vol 1, 117-124.
6. Chalko, T., Gershkovich, V. & Haritos, N. “The Direct Simultaneous Modal Approximation method”, Proc.
IMAC-XIV, Dearborn, 1996, Vol 2, 1130-1136
7. Haritos, N. & Aberle, M., “Using traffic excitation to establish modal properties of bridges”, Proc.
MODSIM'95 Conference, Newcastle, Aust., 1995, Vol. 1, 243-248.
8. Bayissa, W.L., Haritos, N., & Thelandersson, S., “Vibration-based structural damage identification using
wavelet transform”, Mechanical Systems and Signal Processing, Vol. 22, 2008, 1194-1215
9. Bayissa, W.L., & Haritos, N., “Structural damage identification using a global damage identification
technique”, International Journal of Structural Stability and Dynamics, Vol. 9, No. 4, 2009, 745-763
10. Bayissa, W. L. & Haritos, N., “Damage identification in plate-like structures using bending moment
response power spectral density”, Structural Health Monitoring, 6(1), 2007, 5 - 24
11. Perera, R., and Ruiz, A., “A multistage FE updating procedure for damage identification in large-scale
structures based on multiobjective evolutionary optimization”, Mechanical Systems and Signal Processing,
Vol. 22, 2008, 970–991.
12. Cole H. A., “On-Line Failure detection and damping measurement of aerospace structures by Random
Decrement signatures”, NASA Contractor Report (NASA CR-2205), 1973, Washington, 75 p.
13. http://www.gcdataconcepts.com/, last accessed 20th August, 2010.
14. Haritos, N. & James, K., “Dynamic response characteristics of urban trees”, Australian Earthquake
Engineering Conference, Ballarat, Victoria, 21-23 Nov 2008, 10p.
15. Abu-Aisheh, E. & Haritos, N., “Ambient modal testing of bridges using ARMAV Models”, Proc Intl. Conf.
on Mechanics of Structures, Materials and Systems, Wollongong, February, 1999, 75-82.
16. ARTeMIS Extractor, Structural Vibration Solutions A/S, Aalborg, Denmark.
17. Haritos, N. “The application of Experimental Modal Analysis testing to the identification of the in-service
condition of bridge superstructures”, Proc. APSBLAF'96, Monash University, Vol 3, 1996, 91-100.
18. Haritos, N., “Application of vibration-based testing techniques to structural system identification studies”,
Proc. First International Conference on Modern Design, Construction and Maintenance of Structures,
Hanoi, Vietnam, Dec. 2007, Construction Publishing House (Hanoi), 447-454.
19. Haritos, N. & Hewitt, M. “Condition assessment of Sale Swing Bridge using dynamic testing”, Proc. ASEC
2005, Newcastle, Australia, Sept. 2005, 10p.
20. Haritos, N. & Hewitt, M., “Dynamic testing of the Sale Swing Bridge - before and after deck replacement
works”, Proc. ACMSM19, Christchurch, 29th Nov. - 1st Dec., 2006, in Progress in Mechanics of Structures
and Materials, (edit Moss P. J. and Dhakal, R. P.), Balkema, 311-318
21. Haritos, N., Nguyen, C. & Ngo, T., “Wind induced excitation of box girder gantry frames”, accepted for
presentation ACMSM21 Conference, Melbourne, Australia, 8-10 Dec., 2010.