Structural health continuous monitoring of buildings

– A modal parameters identification system

Mirco Muttillo Luca Di Battista
Dept.of Industrial and Information Engineering and Economics Dept. of Information Engineering, Computer Science and
(DIIIE) Mathematics (DISIM)
University of L’Aquila Univerity of L’Aquila
L’Aquila, Italy L’Aquila, Italy
[email protected] [email protected]

Tullio de Rubeis Iole Nardi

Dept.of Industrial and Information Engineering and Economics Energy Efficiency Unit Department (DUEE)
(DIIIE) ENEA Casaccia Research Center
University of L’Aquila Santa Maria di Galeria, Italy
L’Aquila, Italy [email protected]
[email protected]

Abstract— Monitoring systems play a key role in In the field of structural monitoring, excluding some types
maintaining the buildings’ structural health. Although in the of public buildings and private houses, all structures are
last decades the structural monitoring has experienced a unique. This means that there is no common repeatable
considerable growth, the monitoring systems still require solution in all situations. Nevertheless, most of the monitoring
remarkable installation efforts and significant costs. Due to systems for infrastructures and civil structures need long-term
these disadvantages, the spread of such systems was scarce, and monitoring campaigns to evaluate the health of the facility. In
the duration of experimental phases was often short. The aim of the case of extreme events such as earthquakes or explosions,
this work is the design of a Structural Health Monitoring (SHM) monitoring is used to check the condition of the structure and
system to continuously monitor and check the structural
subsequently to provide information on its integrity.
behavior throughout the buildings’ lifespan. The system, made
up of a customized datalogger and slave devices, allows the Ross and Matthews [1] have identified the cases in which
continuous monitoring of structures’ acceleration thanks to its monitoring is required, like the case of changes of existing
ease of installation and low cost. The proposed system is mainly structures, structures subject to external works, during
based on a microcontroller that: i) communicates with the nodes demolitions, structures subject to movement and degradation
via RS485 bus, ii) synchronizes the acquisition samples, iii) of materials and new construction systems. Chondros e
acquires the data measured by the nodes. The system was tested Dimarogonas [2] used The Hu-Washizu-Barr variational
on a cantilever aluminum structure, through three different
formulation for developing the differential equation and the
experimental campaigns and the measured data, collected in an
internal memory of the datalogger, were post-processed via
boundary conditions for the cracked beam as one-dimensional
Matlab algorithm. The results allowed to evaluate the modal continuum. Qiao [3] detects structural damages by applying a
parameters (frequencies, damping and modal shapes) of the signal-based pattern recognition method with a limited or
analyzed structure and its health. single input and output signals. This technique is based on the
sensitive features extraction under excitation of the structural
Keywords— Structural health monitoring (SHM), modal behavior. Srinivasarao [4] have presented a method in beam
parameters, stochastic subspace identification (SSI), data structures for crack identification analyzing the fundamental
acquisition system, health diagnosis mode of cracked cantilever beam by continuous transform. El-
Kafrawy [5] considered the flexural vibration of a cantilever
I. INTRODUCTION beam, fixed-roller beam, and portal frame with a transverse
The applications of structural monitoring concern various surface crack. The natural frequency and the mode shape, in
disciplines such as aerospace and mechanical engineering, but steel beams, change with transverse crack.
in recent decades it is also moving in the field of civil Alampalli [6] focused most studies on correlation modal
engineering and infrastructure. Indeed, these structures parameters changes in structural condition, and results
undergo checks, monitoring and maintenance programs indicate that modal frequencies in conjunction with mode
required by law. The effectiveness of such maintenance and shape may be used to identify the existence of bridge damage
inspections is proven if there is the possibility of promptly and or deterioration of interest. Cross [7] outlines the monitoring
accurately establishing the level of performance. The main campaign of the Tamar Suspension Bridge as well as analysis
sectors of structural monitoring are the hydrocarbon industry, carried out to understand the normal condition of structure and
the companies in the field of dams and highways. These development of a structural health monitoring system. In a
structures are affected by great attention and research efforts data-based approach to a Structural Health Monitoring
are concentrated on them. Residential and commercial (SHM), the lack of data from damaged structures forces the
facilities have received relatively little attention due to the lack use of new techniques for diagnosis. Therefore, the harmful
of awareness of the owners on the benefits and on the variations in the operating or environmental conditions of the
importance of structural monitoring. Therefore, in these cases, structure should be careffully managed, in order to avoid false
efforts are needed to inform owners about the benefits of using alarms. Recent developments in the field of robust regression
the monitoring also through legislative instruments or can provide ways to explore and visualize SHM data as a
subsidize. means of understanding the different sources of outliers.
Indeed, Dervilis [8] illustrated the use of robust regression for
the analysis of SHM data through monitoring from Z24 and Covariance Driven [17] implemented on Matlab environment
Tamar Bridge. are used for modal identification [18-19]. The efficiency of
algorithms has been validated on pseudo experimental data
The first approach to the problem of automatic provided by different numerical models, giving a perfect
identification of modal parameters has been presented by match in terms of frequency, modal shapes and damping with
Verboven [9]. Indeed, with the rapid development and respect to the expected ones, thus proving its robustness. The
reliability of modal analysis techniques under operating proposed system has been tested on cantilever structure
conditions, there has been an increase in attention on this topic anchored to a wooden beam and locked with screws. Results
which led to the formulation of different methods for allowed to characterize the instrument and the sample
automatic identification and monitoring of modal parameters. structure also in relation to the identification of modal
The classical methods of modal identification are applied to parameters.
the experimental data, the order of the model is generally
overestimated, as well as the identification of all the physical II. MATERIALS AND METHODOLOGY
modes present in the band of frequencies of interest. Though
on the new construction, with certain basic assumptions, is Measurements were made on a cantilevered structure
present a numerical model that can give us reliable results, anchored to a wooden beam and secured with screws. In the
once the structure has been realized, it is not always easy to test the cantilever structure stood on the ground for
find the assumptions made during the modeling phase. identification the modal parameters. The test was divided into
three distinct phases. The first phase consisted in choosing the
Moreover, the distance between the numerical model and sampling frequency and the duration of the test. In the second
the real physical model is even more pronounced if we the system was powered, therefore samples were acquired and
consider an existing structure that may not have specific data were saved on a SD memory. For this phase the test is
design indications and from which the behavior predicted by repeated three times for identification. In the last phase the
the numerical model can deviate much from the real behavior post-processing of the data with a Matlab algorithm has been
of the structure. done. The block scheme of the test operating phases is shown
in Fig. 1.
Therefore, in recent years, the experimentations on the
structures and the specific dynamic experiments, allow us to III. GENERAL SYSTEM OVERVIEW
identify the real parameters characterizing the real physical
system which then assume a fundamental character in the The systems that are used for structural monitoring, are
dynamic response of the structure. The main reason for usually based on datalogger and high resolution analog
interest in experimental modal analysis is linked to the
consideration that the dynamic behavior of a structure is an
intrinsic property of the system. Indeed, it depends only on its
physical characteristics and not on the external dynamic
forcing applied. It is interesting to underline that if internal
modifications do not occur (for example building structural
damage), the behavior of the structure remains almost
unchanged over time; if not the analysis of the building
immediately after the damage can be revealed by a variation
of the frequencies and the proper mode of vibrating.
Notwithstanding the growth of the techniques for structure
modeling, the acquisition systems must be precise and
reliable, and this makes them very expensive. Indeed, a
structural monitoring of static parameters such as inclinations,
temperatures and humidity is normally carried out [10-11].
The introduction of new sensors that communicate via
wireless [12] or wired (RS485, Powerline[13]) which
unfortunately give little useful information to identify the
wellness of the structure, they are commonly used for static
monitoring
In this paper, an automatic indoor monitoring system for
structural health is presented. The system is based on a
microcontroller Sam3X8E ARM cortex-M3 [14] with digital
accelerometer, type ADXL355 [15], that allows to measure
main parameters for characterizing the structures. The data
returned by the system are subsequently processed. The
developed system can therefore be used for structural health
monitoring and for dynamic identification of structures. This
device can be integrated with other types of building
monitoring and actuators systems such as energy monitoring,
daylight and occupancy control system [16]. An application
through the identification of modal parameters (frequencies,
damping and modal shapes) of a cantilever structure is Fig. 1. Identification test operating phases
proposed. Furthermore, algorithms based on the SSI-
 accelerometric sensors. Often the high cost due to high
resolution analogue sensors, conditioning and A/D conversion
is unjustified with respect to the performances required to the
system for an effective and cost sustainable activity of SHM
on less important structure.
Today, with the advancement of technologies in the field
of integrated electronics, more precise digital sensors are
emerging. The use of digital accelerometers, which inside
have everything necessary to overcome the problems of
analog outputs, allows to create new systems for structural
monitoring decreasing the costs. In particular, the digital
triaxial accelerometer, ADXL355 of Analog Device, is used
for this system. These sensors communicate with
microcontroller through SPI/I2C interface.
For the prototyping phase, an evaluation board based on
microcontroller Sam3X8E ARM Cortex-M3 is used for node Fig. 2. Identification of modal parameters testing structure.
architecture. For this system, an external power supply circuit
is needed. Moreover, an RS485 communication circuit has
been used. Indeed, this circuit allows to communicate with the
master device and to transfer acquired data. An SD external
memory has been provided so that there is no data loss.
On each node two accelerometers are used. Indeed, thanks
to I2C communication, more than one ADXL355 can be
connected to the same bus using a different address, chosen by
external hardware settings.
The total current consumption of node is 100 mA, that
makes it a Low-power system. Therefore, the nodes can also
be supplied with alternative sources such as photovoltaic Fig. 3. Identification of modal parameters testing set up
panels, energy harvesting circuits [20-23] to recharge
batteries. Thanks to this peculiarity, it is possible to mount the
nodes even at great distances and in environments where there V. MEASUREMENT RESULTS
is no electrical network. The first test of identification of modal parameters shows
how the system manages to get out them. The measurements
IV. EXPERIMENTAL SET UP are made up by six axes. In Fig. 4 an acceleration
The proposed monitoring system has been verified in measurement relative to an axis located on the structure is
laboratory building on the testing structure. On which shown. The post-processing step allows to select modal
identification of modal characteristic has been applied. The parameters from stabilization diagram; usually, in a
cantilever structure (aluminum bar) for the identification of stabilization Diagram, the search for stable modes is made
modal parameters (frequency, damping and modal shapes) has based on the application of the stability criteria. In Fig. 5 the
been anchored to a wooden beam and locked with. The two Stabilization Diagram plot obtained from the data sets
accelerometers of the acquisition node have been put on the performing the SSI – Covariance algorithm.
aluminum bar. The first accelerometer has been mounted at
the end of the bar and the second was positioned at 16.6 cm
from the blocking point. The identification of modal
parameters testing structure is shown in Fig. 2.
The acquisition system was composed by a node and a
master that communicated through RS485 bus. The node
acquired the accelerometers data, saved them on SD and then
transmited to the master device. Moreover, an external power
supply for node and master was required. The complete testing
system is shown in Fig. 3.
The sampling frequency of 250 Hz has been adopted. The
modal parameters identification has been performed through
an algorithm based on the SSI-Covariance Driven
implemented on Matlab and is used on three measures lasting Fig. 4. One axes acquiring through system.
15 minutes each.
 [3] L. Qiao, A. Esmaeily and H. Melhem, "Structural Damage Detection
Using Signal Pattern-Recognition", Key Engineering Materials, vol.
400-402, pp. 465-470, 2008.
[4] D. Srinivasarao, K. M. Rao, G. V. Raju. “Crack identification on a
beam by vibration measurement and wavelet analysis”. International
Journal of Engineering Science and Technology, 2010, 2 (5), 907-912.
[5] A. El-Kafrawy, "Crack detection by modal analysis in 3D beams based
on FEM", International Journal of Mechanics and Materials in Design,
vol. 7, no. 4, pp. 265-282, 2011.
[6] S. Alampalli, " Effects of testing, analysis, damage, and environment
on modal parameters", Mechanical Systems and Signal Processing, vol.
14, no. 1, pp. 63-74, 2000.
[7] E. Cross, K. Koo, J. Brownjohn and K. Worden, "Long-term
monitoring and data analysis of the Tamar Bridge", Mechanical
Fig. 5. Stabilization Diagram plot Systems and Signal Processing, vol. 35, no. 1-2, pp. 16-34, 2013.
[8] N. Dervilis, K. Worden and E. Cross, "On robust regression analysis as
In total, 8 modes, are extracted from operational modal a means of exploring environmental and operational conditions for
analysis. Table I shows the experimentally determined modal SHM data", Journal of Sound and Vibration, vol. 347, pp. 279-296,
parameters. 2015.
[9] P. Verboven, E. Parloo, P. Guillaume and M. Van Overmeire,
TABLE I. ESTIMATED NATURAL FREQUENCIES AND DAMPING "Autonomous structural health monitoring—part I: modal parameter
RATIOS. estimation and tracking", Mechanical Systems and Signal Processing,
vol. 16, no. 4, pp. 637-657, 2002.
Mode n. Frequency [Hz] Damping [%] [10] L. Pantoli, M. Muttillo, G. Ferri, V. Stornelli, R. Alaggio, D. Vettori,
1 4,69 0,42 L. Chinzari and F. Chinzari, "Electronic System for Structural and
Environmental Building Monitoring", Lecture Notes in Electrical
2 25,57 1,72 Engineering, pp. 481-488, 2019.
[11] T. Rubeis, I. Nardi and M. Muttillo, "Development of a low-cost
3 29,29 0,72
temperature data monitoring. An upgrade for hot box apparatus",
4 34,60 0,66 Journal of Physics: Conference Series, vol. 923, p. 012039, 2017.
[12] T. Gabriele, L. Pantoli, V. Stornelli, D. Chiulli and M. Muttillo, "Smart
5 41,59 0,45 power management system for home appliances and wellness based on
wireless sensors network and mobile technology," 2015 XVIII AISEM
6 45,93 0,39 Annual Conference, Trento, 2015, pp. 1-4.
7 56,57 0,82 [13] L. Pantoli, M. Muttillo, V. Stornelli, G. Ferri and T. Gabriele, "A Low
Cost Flexible Power Line Communication System", Lecture Notes in
8 111,19 0,65 Electrical Engineering, pp. 413-420, 2017.
[14] "ATSAM3X8E - 32-bit SAM Microcontrollers", [Online]. Available:
http://www.microchip.com/wwwproducts/en/ATsam3x8e. [Accessed:
19- Mar- 2019].
Having only two sensors available and therefore two
[15] "ADXL355 Datasheet and Product Info | Analog Devices", [Online].
measurement points, only few modes are clearly Available:
distinguishable. http://www.analog.com/en/products/mems/accelerometers/adxl355.ht
ml. [Accessed: 19- Mar- 2019].
VI. CONCLUSION [16] T. de Rubeis, M. Muttillo, L. Pantoli, I. Nardi, I. Leone, V. Stornelli
In this paper, an automatic indoor monitoring system for and D. Ambrosini, "A first approach to universal daylight and
occupancy control system for any lamps: Simulated case in an
structural health based on microcontroller Sam3X8E ARM academic classroom", Energy and Buildings, vol. 152, pp. 24-39, 2017.
cortex-M3 and digital accelerometer ADXL355 is proposed.
[17] R. Brincker and P. Andersen, “Understanding stochastic subspace
This system allows to measure main parameters for identification”, Proceedings of the 24th IMAC, St. Louis, Missouri,
characterizing structures. Moreover, it allows to acquire 2006, pp.279-311.
several data with high resolution for identification the modal [18] A. Yan and J. Golinval, "Null subspace-based damage detection of
parameters. The high number of samples that can be acquired structures using vibration measurements", Mechanical Systems and
and the high synchronization between the nodes ensures Signal Processing, vol. 20, no. 3, pp. 611-626, 2006.
reliable results. The choice of digital accelerometers, [19] M. Döhler, L. Mevel and F. Hille, "Subspace-based damage detection
moreover, allows to eliminate typical problems that arise with under changes in the ambient excitation statistics", Mechanical
Systems and Signal Processing, vol. 45, no. 1, pp. 207-224, 2014.
traditional analog systems, and furthermore it makes the
[20] P. Di Marco, V. Stornelli, G. Ferri, L. Pantoli and A. Leoni, "Dual band
system cheaper. harvester architecture for autonomous remote sensors", Sensors and
The test was carried out to evaluate the natural frequencies Actuators A: Physical, vol. 247, pp. 598-603, 2016.
and the damping ratios of a sample structure. Therefore, to [21] A. Leoni, L. Pantoli, V. Stornelli, G. Ferri, M. Russo and P. Solic,
"90/900 MHz IC architecture for autonomous systems," 2017 2nd
implement a structural health monitoring system, it is International Multidisciplinary Conference on Computer and Energy
important to detect any damage starting from modal Science (SpliTech), Split, 2017, pp. 1-4.
identification. The future development of the proposed system [22] L. Pantoli, A. Leoni, V. Stornelli and G. Ferri, "An IC architecture for
will concern the identification of any damage to the structures. RF Energy Harvesting systems", Journal of Communications Software
and Systems, vol. 13, no. 2, p. 96, 2017.
REFERENCES [23] L. Pantoli, A. Leoni, V. Stornelli and G. Ferri, "Energy harvester for
[1] R. M. Moss, S. L. Matthews, “In-service structural monitoring. a state remote sensors systems," 2016 International Multidisciplinary
of the art review,” Structural Engineer, 1995, vol. 73, pp.23-31. Conference on Computer and Energy Science (SpliTech), Split, 2016,
pp. 1-3
[2] T. Chondros and A. Dimarogonas, "Vibration of a Cracked Cantilever
Beam", Journal of Vibration and Acoustics, vol. 120, no. 3, p. 742,
1998.