Journal of Constructional Steel Research 74 (2012) 58–62

Contents lists available at SciVerse ScienceDirect

Journal of Constructional Steel Research

Dynamic monitoring and numerical modelling of communication towers with FBGbased accelerometers

Paulo Antunes a,b, Rui Travanca c,⁎, Humberto Varum c, Paulo André a,b

a Instituto de Telecomunicações, Aveiro, Portugalb Physics Department, University of Aveiro, Portugalc Civil Engineering Department, University of Aveiro, Portugal

⁎ Corresponding author. Tel.: +351 234370049; fax:E-mail address: rui.travanca@ua.pt (R. Travanca).

0143-974X/$ – see front matter © 2012 Elsevier Ltd. Aldoi:10.1016/j.jcsr.2012.02.006

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 September 2011Accepted 11 February 2012Available online 22 March 2012

Keywords:Steel towersStructural health monitoringFBG based accelerometersNumerical simulationModel calibration

This study presents the dynamic monitoring of two telecommunication tall slender steel towers with anoptical FBG accelerometer. Numerical simulation for both towers was used recurring to finite elementsmodelling in order to demonstrate the feasibility of using optical technology in this type of structural moni-toring. The results show a good agreement between experimental and simulated data, demonstrating thatthe optical accelerometer can be a very useful tool in the monitoring of tall slender structures.

© 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Towers are one of the most important physical supports for theinstallation of radio equipment used for the emission of electromag-netic waves that allow various services, such as radio, televisionand/or mobile communications. Unfortunately, the number of anom-alies observed in such structures is high when compared with otherstructures with equal economic and social importance. A great num-ber of the anomalies observed are due to poor design, ensuing inunsafe structures that can even conduce to its full collapse [1–5]. Par-ticularly, for economic and functional reasons of their own nature,steel towers are lightweight structures with high slenderness andgreat flexibility. The increasingly strength of the materials used inits construction and, consequently, the changes verified in stiffness,mass and damping should lead to a new design approach of thewind action for these type of structures. The resonant response be-comes important when these structures have the first natural fre-quency below 1 Hz. Therefore, dynamic analysis is needed todetermine the resonance response that could be significant comparedwith the background response.

The dynamic analysis of tall slender towers is commonly per-formed in the frequency domain, based on the frequency dependentcharacter of both the wind loads and the mechanical properties ofthe structure. In this approach, gust winds are characterized in aprobabilistic basis, i.e., using statistically derived descriptions of the

+351 234370094.

l rights reserved.

relevant properties, such as frequency content and spatial organiza-tion [2,5]. Nevertheless, more simplified procedures can be used toobtain the structural response. For example, RSA [6] prescribes a sim-plified static analysis that only applies if the first natural frequency isabove 0.5 Hz. Within this method, the first natural frequency is usedonly to check the domain of application of the methodology and tovalidate this simplified procedure. According to the Eurocodes [7,8],simplified quasi-static design procedures can be adopted using theappropriate gust response factors. These factors depend on severalparameters, including the first natural frequency of the tower, itsdamping and the characteristics of the wind. So, also in the design ap-proach followed in the Eurocodes, the first natural frequency is un-doubtedly a key parameter to estimate the response of the structure.

Structural Health Monitoring (SHM) is essential to access thestructural integrity and guaranty the lifetime of such structures. Akey parameter to be monitored is acceleration fromwhich the naturalfrequencies of the structure can be obtained. The changes verified innatural frequencies can be related with degradation of the structureand this parameter is a very good indicator of the structure health,allowing taking action preventively if needed, saving money and,sometimes, lives. Occasionally, conventional electronic accelerome-ters can be used. However, the high level of electromagnetic radiationnear the antennas can easily mislead to wrong results and can eveninterfere with radio operation with inevitable economic losses forthe operator. Another approach is use all-optical instrumentationlike the one proposed in this paper. The accelerometer used in thiswork is based on Fiber Bragg Gratings (FBG) technology. So, theusage of this technology presents many advantages, particularly:a) low cost, b) immunity to electromagnetic interference, c) do not

a) TWR-01 b) TWR-02

Fig. 1. Photograph of the towers used for this study.

a) b)

Fig. 3. a) Photography of the TWR-01 and b) directions for which the accelerometer issensitive (obtained through Google maps in 29 March 2011).

59P. Antunes et al. / Journal of Constructional Steel Research 74 (2012) 58–62

interfere in the proper working of antennas because does not emit ra-diation, d) possibility of multiplex a large number of sensors in thesame optical fiber (acceleration, velocity, strain, relative displace-ment, pressure, temperature, etc.) reducing the need of heavy ca-bling, e) possibility of remote monitoring since the fiber presentvery low attenuation, and f) suitable to be embedded into structureswith low visual impact and/or without damage due to the low sizeand volume [9–11].

2. Description of the towers

Tower typology varies widely across countries, because of usagepurpose and actions to be considered in structural analysis and de-sign. For this study were chosen two distinct tall slender monopoleswith 50 m high installed in Portugal (Fig. 1).

According to the design elements supplied, the first monopole(TWR-01) is formed by five different steel modules with tubular crosssection joined by bolted connections. This monopole is fixed to a rein-forced concrete foundation block with the dimensions 4.00×4.00 m inplan and 2.25 m in deep. The secondmonopole (TWR-02) is constituted

a) TWR-01 b) TWR-02

Fig. 2. Photography of the accelerometers assembly.

by seven different modules with hexadecagonal cross section with thecross section dimensions varying in height. The various modules arelinked by forced fit that is achieved with pre-strength steel wire ropesand, consequently, with overlap of the different steel modules. Accord-ing to the design elements, the secondmonopole is fixed to a reinforcedconcrete semi-deep foundation block with dimensions 3.30×3.30 m inplan and 3.60 m in deep. Apart from the minor differences in terms ofthe geometry of the steel monopoles observed in-situ, it were detecteddifferences between the drawings supplied and the real geometry of thefoundations, namely in terms of the geometry in plan. Also, elements forthe soil foundation characterization were not available.

3. Dynamic monitoring and numerical simulation

3.1. Dynamic monitoring

The used acceleration sensor consists of two FBG pairs, each pairperpendicularly placed to the other. Each FBG pair is responsible formeasuring acceleration in one direction. For this propose, the FBGs arefixed between an inertial mass and a fixed support. For each pair, oneFBG is placed in one side of the mass and the other on the other side.In this way, when the inertial mass moves during acceleration, oneFBG will contract and the other expand. This technique allows

a) b)

Fig. 4. a) Photography of the TWR-02 and b) directions for which the accelerometer issensitive (obtained through Bing Maps on 7th March 2011).

a) TWR-01 b) TWR-02

Fig. 5. Accelerograms recorded.

a) TWR-01 b) TWR-02

Fig. 6. Tower frequencies' spectra.

Table 3

60 P. Antunes et al. / Journal of Constructional Steel Research 74 (2012) 58–62

improving sensitivity and simultaneously makes the accelerometertemperature insensitive [12].

The optical FBG based bi-axial accelerometer was fixed on a rigidsteel base plate and mounted at the top of the tower near antennas.For TWR-01, the optical FBG based bi-axial accelerometer was fixedon a rigid steel support system and mounted at the top of the towerwith a bolted connection (Fig. 2-a). For TWR-02, the steel base platewas connected to the monopole using stainless steel strips (Fig. 2-b), avoiding the need of bolt or weld connections which wouldcause difficulties in the works at that altitude and, certainly, causinglocal damage on tower anticorrosion protection. These proceduresmade possible that all testing works were done in less than 3

Table 1Frequencies obtained experimentally for TWR-01 and TWR-02.

Peaknumber

Frequency [Hz]

TWR-01 TWR-02

Direction 1 Direction 2 Direction 1 Direction 2

1 0.429 0.414 0.661 0.6682 1.628 1.555 2.085 2.2163 3.451 3.451 4.599 4.5994 4.170 4.171 4.817 4.817

Table 2Frame sections mechanical properties for TWR-01.

Sectionname

Area I AS S Z

[m²] [m4] [m²] [m³] [m³]

S-0406- 0.007981 0.000160 0.003991 0.000786 0.001016S-0609- 0.014971 0.000678 0.007486 0.002223 0.002867S-0812- 0.020027 0.001622 0.010014 0.003991 0.005131S-1016- 0.030102 0.003812 0.015052 0.007504 0.009644S-1219- 0.036179 0.006618 0.018090 0.010857 0.013931

hours for each tower, including climbing, assembling, setup installa-tion, measuring, disassembling and descend of men and equipment.Mechanical impulses were applied in the mid section of the struc-tures during which the acceleration data was measured in twoindependent horizontal directions, i.e., parallel to the ground. Theacceleration data was acquired during 105 s using an I-MONE-USB Interrogation Monitor Unit from Ibsen, at a sampling rate of950 Hz. In Figs. 3 and 4 are shown towers location and the measure-ment directions.

Frame sections mechanical properties for TWR-02.

Sectionname

Area I AS S Z

[m²] [m4] [m²] [m³] [m³]

S-0400-4 0.005059 0.000100 0.002960 0.000502 0.021112S-0620-4 0.007906 0.000380 0.004601 0.001226 0.079461S-0620-5 0.009794 0.000469 0.006142 0.001513 0.079091S-1005-5 0.015924 0.002017 0.009981 0.004013 0.339915S-1005-6 0.018995 0.002401 0.011916 0.004778 0.338937S-1184-6 0.022569 0.003967 0.014101 0.006700 0.555754S-1184-8 0.028602 0.005013 0.019276 0.008468 0.553490S-1775-8 0.045012 0.017805 0.029497 0.020056 1.875663

Table 4Mass distribution for TWR-01.

Elements Mass %

[kg]

Steel 8576 77.8%Steel rings 223 2.0%Ladder and feeders 1274 11.5%Work platforms 591 5.4%Antennas 194 1.8%Workers on tower 163 1.5%

Table 5Mass distribution for TWR-02.

Elements Mass %

[kg]

Steel 9092 69.1%Overlap 1861 14.1%Ladder and feeders 1274 9.7%Work platforms 561 4.3%Antennas 204 1.6%Workers on tower 163 1.2%

Table 6Modal frequencies and participating mass ratios for TWR-01.

Stepnumber

Frequency Mass ratios

[Hz] Individual Cumulative

1 0.464 0.884 0.8842 1.719 0.092 0.9773 4.016 0.016 0.9924 7.740 0.005 0.9975 12.978 0.002 0.998

61P. Antunes et al. / Journal of Constructional Steel Research 74 (2012) 58–62

The time history of the acceleration response during the applica-tion of mechanical impulses is presented in Fig. 5-a) and b) forTWR-01 and TWR-02, respectively, and for both measured directions.From the time history of the acceleration response it is possible to ob-tain the eigenfrequencies of the structures by Fast Fourier Transform(FFT). The eigenfrequencies are obtained from the frequencies spectraby peak picking. For both towers, the frequencies spectra calculatedare presented in Fig. 6. In Table 1 is presented the frequencies indicat-ed in Fig. 6-a) and b) and obtained experimentally with the opticalaccelerometer for both measurable directions.

3.2. Numerical modelling

Due to the uncertainties explained above and particularly to theones related with the foundations geometry and soil characteristics,the simulation models adopted in this work are a simple cantilever,

a) TWR-01: f1=0.464 Hz b) TWR-01: f2=

d) TWR-02: f1=0.719 Hz e) TWR-02: f2=

Fig. 7. Modal shapes for T

i.e., without considering the soil–structure interaction effect, andwith the respective geometrical characteristics of the cross sectionsand mechanical properties of the materials. The mechanical proper-ties of frame sections are presented in Tables 2 and 3. All discreteand distributed masses related with structural and non-structural el-ements, such as platforms, ladders, feeders and antennas, wereaccounted. A more refined mass distribution was adopted accountingfor each element supported in the structure, which are normallyneglected in the design phase, e.g., steel rings and overlap of steelmodules. These particularities were modelled throughout the consid-eration of discrete masses at each position. The mass distributionsused for simulations of TWR-01 and TWR-02 are shown in Tables 4and 5, respectively. All simulation models used for this study are bi-dimensional, i.e., with symmetric geometry and masses in plan. Thestiffness matrix and frequencies were numerically obtained usingthe Finite Element Method (FEM) and the characteristic equation re-spectively. The software used for these simulations was SAP2000

1.719 Hz c) TWR-01: f3=4.016 Hz

2.287 Hz f) TWR-02: f3=5.377 Hz

WR-01 and TWR-02.

Table 7Modal frequencies and participating mass ratios for TWR-02.

Stepnumber

Frequency Mass ratios

[Hz] Individual Cumulative

1 0.719 0.850 0.8502 2.287 0.117 0.9673 5.377 0.023 0.9904 10.337 0.005 0.9965 16.459 0.002 0.998

Table 9Comparison between experimental and simulation results in terms of frequencies forTWR-02.

Stepnumber

Experimental [Hz] Simulation Relative error

Direction 1 Direction 2 [Hz] Direction 1 Direction 2

1 0.661 0.668 0.719 8.8% 7.6%2 2.085 2.216 2.287 9.7% 3.2%3 4.599 4.599 5.377 16.9% 16.9%

62 P. Antunes et al. / Journal of Constructional Steel Research 74 (2012) 58–62

from Computers and Structures, Inc. The modal shapes for the threefirst modes are presented in Fig. 7 and in Tables 6 and 7 are shownthe natural frequencies and the participating modal mass ratiosvalues obtained for the first five modes.

3.3. Discussion

The comparisons between experimental and simulated values of thetowers natural frequencies are presented in Tables 8 and 9 with theindication of the relative error. Themeasured values of natural frequen-cies obtained with the FBG biaxial accelerometer are values expectedfor this type of structures. More, measured values are consistent withthe results obtained with the numerical models presented. ForTWR-01, it was obtained a difference between the measured first fre-quency (0.414 Hz) and the simulated value (0.464 Hz) of 12.1%. ForTWR-02, the difference between the measured (0.661 Hz) and the sim-ulated first natural frequency (0.719 Hz) is 8.8%. For the second mode,with more than 95% participating mass obtained in the simulation, thedifferences between measured and simulated frequencies values are10.5% and 9.7%, for TWR-01 and TWR-02, respectively. For mode 3and above, the participatingmass is reduced so these modes can be dis-carded in this comparative analysis. The discrepancies obtained are jus-tified by the mass distribution adopted in the numerical models and/oroverestimation of the stiffness in the numerical models. For example, inreal in-situ conditions: a)mass distribution do not correspond preciselyto the distribution adopted in the numerical model, b) a stiffness losscan occur due to connections, c) degradation of existing materials mayalso occur and this is not considered in models, and d) soil–structureinteractions can play a very important role and it was not considereddue to the reasons presented.

4. Conclusions

The dynamic analysis of towers is commonly performed in the fre-quency domain, considering the frequency dependent character ofboth the wind loads and the properties of the structure. Nevertheless,simplified procedures can be used. However, in recently publishedstandards, as the Eurocodes, the natural frequencies, in particular,the first natural frequency is assumed as a key parameter to deter-mine the structural response, to assess the structural safety of exist-ing structures and to design new structures. In this study it wasdemonstrated the feasibility of dynamic monitoring of tall slenderstructures with optical technology, namely with optical FBG acceler-ometers. Experimental values were compared with results obtainedwith finite elements analysis. The first natural frequency of the

Table 8Comparison between experimental and simulation results in terms of frequencies forTWR-01.

Stepnumber

Experimental [Hz] Simulation Relative error

Direction 1 Direction 2 [Hz] Direction 1 Direction 2

1 0.429 0.414 0.464 8.2% 12.1%2 1.628 1.555 1.719 5.6% 10.5%3 3.451 3.451 4.017 16.4% 16.4%

monitored structures was well predicted by the simulation modelwith relative errors of 12.1% and 8.8%, for TWR-01 and TWR-02, re-spectively. The frequencies of the towers for the second vibrationmode were also accurately predicted with relative errors of 10.5%and 9.7%. The discrepancies obtained are justified by the mass distri-bution adopted and/or overestimation of the stiffness in the numeri-cal models. In this work, errors in the mass distributions wereminimized but it was not possible to minimize the eventual differ-ences associated to the stiffness reduction, as explained before, asso-ciated to soil–structure interaction, eventual material deterioration ordifferences in terms connection conditions influencing the local stiff-ness. Still, FBG biaxial accelerometer is presented as an excellent toolfor numerical model calibration, not only for existing structures, butalso in the optimization of new solutions and, certainly, to a betterunderstanding of the response of this type of structures.

Acknowledgements

P.F.C. Antunes acknowledge the financial support from “Fundaçãopara a Ciência e Tecnologia” (FCT) through the PhD fellowship SFRH/BD/41077/2007. The support of “Vodafone Portugal, ComunicaçõesPessoais, S.A.” and “Telcabo, Telecomunicações e Electricidade, Lda.”is most gratefully acknowledged.

References

[1] Smith B. 50 years in the design of towers and masts. From IASS recommendationsto current procedures. International Association for Shell and Spatial Structures(IASS) Symposium; 2009. Valencia.

[2] Smith B. Communication structures. London: Thomas Telford; 2007.[3] Støttrup-Andersen U. Analysis and design of masts and towers. International

Symposium on Lightweight Structures in Civil Engineering; 2002. Warsaw.[4] Støttrup-Andersen U. Masts and towers. International Association for Shell and

Spatial Structures (IASS) Symposium; 2009. Valencia.[5] R. Travanca, Radio communication towers. Pathologies and design, Master Thesis,

University of Aveiro, 2010 (in Portuguese).[6] RSA. Regulamento de Segurança e Acções para Estruturas de Edifícios e Pontes,

Decreto-Lei n.º 235/83 de 31 de Maio; 1983.[7] CEN, European Committee for Standardization. EN 1991-1-4, Eurocode 1: actions

on structures — part 1–4: general actions — wind actions; 2005.[8] CEN, European Committee for Standardization. EN 1993-3-1, Eurocode 3— design

of steel structures, part 3–1: towers, masts and chimneys — towers and masts;2006.

[9] Antunes PFC, et al. Optical fiber accelerometer system for structural dynamicmonitoring. IEEE Sens J 2009;9:1347–54.

[10] Majumder M, et al. Fibre Bragg gratings in structural health monitoring — presentstatus and applications. Sens Actuators A 2008;147:150–64.

[11] Othonos A, Kalli K. Fiber Bragg gratings: fundamentals and applications. Telecom-munications and Sensing: Artech House; 1999.

[12] P. Antunes, et al. Biaxial fiber Bragg grating accelerometer and inclinometer withtemperature insensitivity. IEEE Sensors Journal, in press.