Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 55, 1, pp. 241-252, Warsaw 2017 DOI: 10.15632/jtam-pl.55.1.241 EXPERIMENTAL VALIDATION OF THE USE OF ENERGY TRANSFER RATIO (ETR) FOR DAMAGE DIAGNOSIS OF STEEL-CONCRETE COMPOSITE BEAMS Tomasz Wróblewski, Małgorzata Jarosińska, Małgorzata Abramowicz West Pomeranian University of Technology, Department of Theory of Structure, Szczecin, Poland e-mail: wroblewski@zut.edu.pl, jarosinska@zut.edu.pl, mabramowicz@zut.edu.pl Stefan Berczyński West Pomeranian University of Technology, Institute of Manufacturing Engineering, Szczecin, Poland e-mail: Stefan.Berczynski@zut.edu.pl Thepaper presents changes ofmodal parameters suchas natural frequencies, damping ratios and energy transfer ratios (ETR) following damage of the connection of a steel-concrete composite beam. Sensitivity analysis of the parameters is conducted both for numerical simulation and experimental results. The energy transfer ratio ETR is determined: i) for the whole beam – as a global parameter which can be used for damage detection, ii) for segments of the beam – as a local parameter which can be used for damage localisation. Keywords: damage diagnosis, energy transfer ratio (ETR), steel-concrete composite beams 1. Introduction Steel-concrete composite beams are often used in the bridge engineering as main carrying gir- ders. Safety of bridge structures, both during constructional work and particularly during their use, is of major importance. Therefore, continuous monitoring of the structure condition is an increasingly common practice with a view to damage diagnosis in early stages for substantial reduction of the cost of necessary repairs.Monitoring systems are often based onnon-destructive damage detection methods. They enable damage detection in places inaccessible during visual inspection, e.g. in shear connectors of steel-concrete composite bridges (Liu and deRoeck, 2009; Li et al., 2014a,b; Dilena andMorassi, 2004, 2009). Non-destructive methods are often based on measurements of vibration of the structure. Vibrations that are excited by a changing load are recorded and modal characteristics of the structure are analysed. When damage appears, stiffness, damping and sometimes mass of the structure change, causing changes in frequency response functions. This is reflected in changes of modal parameters, including natural frequencies, mode shapes and damping parameters. Methods based on analysis of such changes can provide an effective tool to damage diagnosis. One of them, which deserves an in-depth analysis, is based on changes of the energy transfer ratio (ETR). Since the method is used for non-proportionally damped structures, it has gained little recognition as most studies assume analysed structures to be proportionally damped. The energy transfer ratio was defined by Liang and Lee (1991). It defines the ratio of energy which is transferred between different mode shapes. Based on an experimental research con- ducted on amodel of a composite bridge, the authors demonstrated that ETRwas much more sensitive to damage than the natural frequency or damping ratio. Their study revealed that modal energy was transferred not only betweenmode shapes but also between their fragments. Itmeans that ETR can be determined both globally, for thewhole construction, and locally, for its segments, which can be used for damage localisation. Similar studies conducted on composite 242 T.Wróblewski et al. bridge models were also carried out by Kong et al. (1996), Liang et al. (1997), Wang and Zong (2002, 2003). The present paper is a continuation of two earlier studies published by Wróblewski et al. (2011, 2013) which demonstrated a method of how ETR changes could be used for damage detection and localisation in steel-concrete composite beams. The results obtained in numerical analysis confirmed that ETR was more sensitive to changes in steel-concrete beam structure than the natural frequency or damping ratio. However, it should be pointed out that the studies were based on numerical damage simulation only. The present paper assesses changes of modal parameters both based onnumerical simulation results and experimental data. ETR is discussed and tested as a damage detection tool, i.e. as a global parameter determined for thewhole beam and as a local parameter determined for segments of the beam, which can be useful for damage localisation. 2. Experimental research The analysed composite beam measures 3200mm in length and consists of IPE 160 made of S235JRG2 steel, connected to a concrete slabmade ofC30/37 concrete. The connection between the concrete and steel ismade usingheaded steel studs, 10mm in diameter and 50mm inheight. The studs are placed in pairs every 150mm.The cross-section of the beamand stud distribution along the beam are shown in Fig. 1. Fig. 1. Composite beam: (a) cross-section, (b) distribution of connectors Experiments have been conducted for a free-free beam.Theywere carried out by suspending the tested beam on a steel frame by means of four elastic steel cables. The places where the cables were attached to the concrete slab overlapped with the nodes of the fundamental flexural mode shape of vibration of the beam. An impulse testwithamodalhammerwas conducted to excite vibrationof the structure.The system responsewasmeasured in three orthogonal directions using piezoelectric accelerometers. LMS SCADAS III, a multi-channel analyser, was used to measure the signal. It was connected to a workstation, fitted with LMSTest.Lab software. A grid of 52 measurement points were defined on the beam. In these points, accelerations were measured. Two additional points, 53 and 54, were situated at one end of the beam. To excite flexural vibration of the beam, force impulse was applied at point 53 perpendicularly to Experimental validation of the use of energy transfer ratio... 243 the slab (−Y direction). To excite axial vibration of the beam, a force impulse was delivered at point 54, on the slab edge (+X direction). A grid of measurement points and the directions of the force impulse are presented in Fig. 2. Fig. 2. Measurement point grid, points and directions of the force impulse The Polymax algorithm, available in Test.Lab, was used to determine parameters of the modalbeammodel.Thepolesof the systemwere identifiedbasedon theanalysis of a stabilisation diagram. Only the measurement points on the beam mid plane (2,4,6,8, . . . ,52) were used in the analysis. This approach substantially made it easier to find flexural vibration modes which were later analysed to detect and locate damage. It also enabled analysis of a two-dimensional beammodel. Natural frequencies fi,exp obtained in experimental research and their corresponding modal damping ratios ξi,exp for the first five flexural mode shapes are presented in Table 1. Table 1.Modal parameters obtained during experimental research i→ 1flex 2flex 3flex 4flex 5flex fi,exp [Hz] 74.68 172.72 273.01 373.40 472.03 ξi,exp [%] 0.15 0.30 0.41 0.39 0.34 In the successive stage of the experiment, connection damage was introduced into the be- am. The damage was obtained by removing concrete around a pair of studs, which eliminated the interaction between steel and concrete (Fig. 3). Dilena and Morassi (2003) used the same approach to simulate damage. Fig. 3. Damage of the connection 244 T.Wróblewski et al. Twodegrees of damage, denotedbyUZ1andUZ2,were introduced into thebeam. Indamage UZ1 concrete was removed around one pair of studs (Fig. 4a) while in damage UZ2 around two pairs of studs (Fig. 4b). Fig. 4. Damage localisation: (a) damage UZ1, (b) damage UZ2 3. Numerical analysis The numerical model of the beam has been created using the Rigid Finite Element method (Kruszewski et al., 1975, 1999;Wittbrodt et al., 2006). The central idea of themethod is division of a real system into rigid bodies which are called rigid finite elements (RFE), which are then connected by means of spring-damping elements (SDE). For continuous parts of a structure, it is customary to start creating a model with segmentation of a beam into equal or nearly equal segments. This segmentation is also called a primary division, see Fig. 5a. A spring-damping element SDE is placed in the centre of gravity of every segment. This SDE is supposed to concentrate all spring and damping properties of a given segment. In Fig. 5b, SDE elements are shown as⊗ symbols. In the 2Dmodel, each SDE ismade up of three (two translational and one rotational) independent spring-damping elements. The next step is to connect SDEs created in the primary division by means of RFE. This is the so-called secondary division, see Fig. 5b. While creating the model, the steel and concrete parts of the beam are treated separately. Owing to this, the elasticity of the connection can be taken account of the simulations. The connectionbetweenadjacent rigidfinite elements (RFE)modelling theconcrete slab isperformed in the classical way, i.e. using one SDE located in the axis of the slab. In the case of the steel beam, another approach is proposed. One SDE is substituted with three separate SDEs placed in the axes of the web and the flanges of the I-section. The SDEs modelling the connection are located in the place where in the real beam a pair of steel studs would be found. Figure 5 presents the composite beammodel. To develop a mathematical model of a composite beam, it is necessary to know parame- ters characterising the I-section, the concrete slab and the connection. Some parameters, e.g. Young’s modulus of the steel, density of the materials, Poisson’s ratios, cross-sectional areas, are determined on the basis of measurements and literature data. Other parameters necessary to define themodel are determined in two-stage parametric identification. The first identification stage has been focused on determining the missing stiffness parame- ters, i.e. Young’smodulus of the reinforced concrete slabEc and stiffness of the connectionKv in the vertical direction (perpendicular to the connection plane) and in the horizontal directionKh (parallel to the connection plane). The stiffness of the connection is influenced by spring proper- ties of the steel studs, stiffness of concrete surrounding the studs and stiffness of the adjacent section of the steel-concrete interface. The first identification criterion is the best fit of the first five frequencies of flexural vibrations determined numerically to those found experimentally. Experimental validation of the use of energy transfer ratio... 245 Fig. 5. RFE model of the composite beam: (a) primary division, (b) secondary division The second identification criterion is defined as the complete fit of the fundamental frequency of axial vibration determined numerically and experimentally (it is due to strong correlation of this frequency with Young’s modulus of the concrete slab Ec). Identification process has been conducted with a MATLAB optimization toolbox, using fmincon function which finds a mi- nimum of a constrained nonlinear multivariable function. The function can be used to find a minimum with four different algorithms. Active-set and Interior-point were tested during the analysis. Finally, the second algorithm was selected. The algorithm required determination of the Hessian matrix, a matrix of second-order derivatives. The Hessian was determined using BFGS, a variable metric method. The obtained fit of frequencies and the results of stiffness parameter identification are presented in Table 2. Table 2. Identification results of stiffness properties of the beam i→ 1flex 2flex 3flex 4flex 5flex 1ax fi,exp [Hz] 74.68 172.72 273.01 373.40 472.03 585.94 fi,num [Hz] 74.93 170.91 275.55 375.06 469,05 585.94 ∆i 0.34% −1.05% 0.93% 0.44% −0.63% 0.00% Ec [N/m 2] 2.89 ·1010 Kh [N/m] 2.19 ·10 8 Kv [N/m] 1.44 ·10 8 At the second identification stage, the values of parameters definingdampingproperties, loss factors for concreteQ−1c and connectionQ −1 con, have been identified.According toRao (2004), the loss factor for steel is Q−1s ∈ 〈2 ·10 −4,6 ·10−4〉. Finally, the loss factor for steel was predefined to be equal Q−1s = 4 · 10 −4. The identification process was based on fitting the amplitudes 246 T.Wróblewski et al. of the frequency response function (FRFs) obtained using RFE model with those obtained during the experiment.ForFRFsdetermined in thevertical direction (axisY , impulse excitation at point 53 : −Y ), the frequency range from 50Hz do 350Hz was fitted. The range includes frequencies of the first threemodes of flexural vibration. For FRFs determined in the horizontal direction (axis X, impulse excitation at point 54 : +X), the frequency range including the fundamental longitudinal vibrationwas considered.The following loss factorsweredetermined in the identification:Q−1c =0.0100 andQ −1 con =0.0269. Figure 6 shows a comparison of illustrative FRFs obtained experimentally with those calculated with the RFE model. The numbers of points from a grid of measurement points used during the experiment and their corresponding rigid finite elements numbers from the numerical model are given in Fig. 6. Location of RFE No. 7 and RFE No. 17 corresponds exactly to the location of point No. 16 and point No. 40 which was placed on the bottom flange of the I-section during experiment (see Fig. 2). Fig. 6. A comparison of FRFs (Y direction) obtained in the experimental research and those simulated with the RFEmodel. The excitation force applied at point 53 :−Y The comparison of the experimental and numerical mode shapes of vibration has been de- termined using theModal Assurance Criterion – MAC (Allemang and Brown, 1982) MAC = |QTnumQexp| 2 (QTnumQnum)(Q T expQexp) (3.1) where the vectorsQnum andQexp denote respectivelymode shapes determined numerically and experimentally. The obtained values of MAC are presented in Table 3. Table 3.Values of the coefficient MAC i→ 1flex 2flex 3flex 4flex 5flex MAC i 0.97 0.92 0.88 0.82 0.72 According to literature reports (Uhl, 1997), the fit of a numerical model and an experiment occurs when the MAC is above 0.8. As can be seen from Table 3, the condition is not satisfied for the fifthmode shape. Therefore, the fifthmode shape is not included in further analysis. Numerical simulation of damage in the connection, which included damage introduced in experiments into the real beam, has been conducted for the developed model. Damage in the connection was simulated by changing the spring properties of SDEs that modelled the connec- tion: the stiffness coefficients of the respective SDEs were preset to equal zeroKh =Kv =0. Experimental validation of the use of energy transfer ratio... 247 4. Analysis results The sensitivity analysis of modal parameters following damage introduced into the beam has begun with examination of changes in the fundamental modal parameters, such as natural frequency and damping ratio. Then, the energy transfer ratio (ETR) was analysed. ETR was determined globally, for the whole beam, and locally, for some parts of the beam. 4.1. Vibration frequency and damping ratio The changes of natural vibrations and damping ratio after damage introduction both for the calculated and experimental data are presented in Tables 4 and 5, respectively. Table 4.Changes of natural frequencies following beam damage Beam state → UZ0 UZ1 UZ2 i ↓ fi,num [Hz] fi,num [Hz] ∆i [%] fi,num [Hz] ∆i [%] N u m er ic a l a n a ly si s 1flex 74.93 74.53 −0.53 73.96 −1.29 2flex 170.91 170.23 −0.40 168.89 −1.18 3flex 275.55 272.87 −0.97 265.71 −3.57 4flex 375.06 369.74 −1.42 363.58 −3.06 i ↓ fi,exp [Hz] fi,exp [Hz] ∆i [%] fi,exp [Hz] ∆i [%] E x p er im en t 1flex 74.68 73.97 −0.95 73.29 −1.86 2flex 172.72 170.22 −1.45 167.63 −2.95 3flex 273.01 267.45 −2.04 259.91 −4.80 4flex 373.40 365.50 −2.12 357.81 −4.18 As can be seen in Table 4, high consistency of natural frequency changes determined experi- mentally and numerically has been achieved. Somewhat larger changes of the parameter, follo- wing damage introduction, were recorded in experimental measurements. In each analysed case, an increase of damage extent caused a decrease in the natural frequency, which was directly associated with a stiffness loss of the beam. Table 5.Changes of the damping ratio following beam damage Beam state → UZ0 UZ1 UZ2 i ↓ ξi,num [%] ξi,num [%] ∆i [%] ξi,num [%] ∆i [%] N u m er ic a l a n a ly si s 1flex 0.26 0.28 9.62 0.33 27.04 2flex 0.41 0.42 4.74 0.51 26.47 3flex 0.51 0.57 12.98 0.86 68.67 4flex 0.61 0.68 11.47 0.78 27.02 i ↓ ξi,exp [%] ξi,exp [%] ∆i [%] ξi,exp [%] ∆i [%] E x p er im en t 1flex 0.15 0.16 4.79 0.16 4.99 2flex 0.30 0.28 −5.56 0.29 −1.86 3flex 0.41 0.44 9.28 0.44 8.27 4flex 0.39 0.44 12.86 0.45 15.66 Changes in the damping ratio determined experimentally and numerically do not show the level of similarity as in the frequency. The ratio is more sensitive than natural frequencies to damage introduction, which is particularly true for numerical simulation results. There is a tendency for the damping ratio to increase as damage extent increases. 248 T.Wróblewski et al. 4.2. Energy transfer ratio ETR The sensitivity analysis of the energy transfer ratio ETR for damage introduced into the steel-concrete composite beam has been conducted in two stages. First, the results of numerical simulations of damage produced with the rigid finite element model were analysed. Then, the sensitivity analysis of ETR based on experimental results was conducted. The equilibrium of a vibrating system with n degrees of freedom, including the effect of damping, can be given by a system of differential equations MẌ+CẊ+KX=F (4.1) where:M,C,K – aremass, damping and stiffness matrices, Ẍ, Ẋ,X – vectors of acceleration, velocity and displacement,F – the external force vector. When considering free damped vibrations, the above equation can be written as Ÿ+CẎ+KY=0 (4.2) where:Y=M0.5X,C=M−0.5CM−0.5,K=M−0.5KM−0.5. For the system defined with Eq. (4.2), there is a set of modal parameters, i.e. natural fre- quencies ωi, modal damping ratios ξi and the mode shapes Pi, where i = 1, . . . ,n. For the generalized stiffness matrix K, there are eigenvectors Qi, for which in proportionally damped systems it is true that Qi =Pi. In non-proportionally damped systems, for which Qi 6=Pi, it is true that (Liang and Lee, 1991) ζi = Im ( 1 2ωi Q T i CPi Q T i Pi ) (4.3) where ζi denotes the energy transfer ratio ETR. The component (Q T i CPi)/(Q T i Pi) is defined as the generalized Rayleigh quotient. Amore comprehensive discussion of ETR can be found in reports by Liang and Lee (1991) and byWang and Zong (2002). 4.2.1. Numerical analysis Table 6 presents changes in the global ETRdetermined in numerical simulation of the beam damage. Table 6.Changes of global ETR following beam damage Beam state → UZ0 UZ1 UZ2 N u m er ic a l a n a ly si s i ↓ ζi,num ·10 2 [%] ζi,num ·10 2 [%] ∆i [%] ζi,num ·10 2 [%] ∆i [%] 1flex 2.9 3.7 27 5.5 89 2flex 4.3 5.0 16 12.3 187 3flex 4.9 7.6 56 28.9 495 4flex 5.4 7.4 38 12.5 132 The analysis of the data demonstrates that the value of ETR increases with an increase in the degree of damage in the connection of the beam. ETR can be said to be substantially more sensitive to this type of damage than both the natural frequency and damping ratio. The largest change of ETR occurred in the thirdmode shape, for UZ2 damage, when it amounted to 495%. As mentioned above, the local ETR was determined for parts of the beam. It is to be emphasised that the division into partswas conducted individually for eachmode shape, so that Experimental validation of the use of energy transfer ratio... 249 each analysed part covered the area of a substantial curvature while at the same time avoiding nodes of the mode shape. The analysis started from the secondmode shape. Figure 7 shows changes of the locally determined ETR following damage introduction. The figure also presents the calculated mode shapes for the undamaged beam – the horizontal axis shows the localisation of damaged stud connectors. Fig. 7. Changes of local ETR ζi,num with mode shapes and damage area: (a) 2ndmode shape, (b) 3rdmode shape, (c) 4th mode shape Thechanges of the locallydeterminedETRdistinctly showtheareaofdamage.Theminimum sensitivity was observed for UZ1 damage, where the maximum change was 181%. Much larger changes occurred for UZ2 damage, with the order of magnitude ranging from several hundreds to even several thousandspercent.Thehighest sensitivitywas observed for the thirdmode shape when ETR change was 1836%. 4.2.2. Experimental analysis To determine ETRbased on experimental data, it is necessary to know the dampingmatrixC. The matrices determined for the RFE model for successive states of the beam, UZ0, UZ1 and UZ2, were used in the analysis. Table 7 shows changes of global ETR determined with the experimental data. The results include changes that occurred for the second, third and fourthmode shape, i.e. thosewhichwere going to be analysed in damage localisation. The ETR changes determined with the experimental data occurred in the range of 13%- -268%. Compared to numerical analysis results, the changes are smaller. However, it is to be 250 T.Wróblewski et al. Table 7.Changes of the global ETR following beam damage Beam state → UZ0 UZ1 UZ2 E x p er im en t i ↓ ζi,num ·10 2 [%] ζi,num ·10 2 [%] ∆i [%] ζi,num ·10 2 [%] ∆i [%] 2flex 20.5 50.8 147 75.4 268 3flex 26.8 7.2 73 68.6 156 4flex 30.6 8.0 74 34.5 13 emphasised that the numerical analysis was not affected by distortions which did take place during real measurements. The presence of uncontrolled distortions and measurement noise may affect the results. However, despite this fact, changes of ETR, compared to frequency and damping ratio changes, are much larger. It demonstrates that ETR is more sensitive to the damage introduced into the beam. Changes of local ETR following the connection damage are presented in a graphical form in Fig. 8. Figure 8 also shows experimentally the determined mode shapes for the undamaged beam – the horizontal axis shows the localisation of damaged stud connectors. Fig. 8. Changes of local ETR ζi,exp with mode shapes and damaged areas: (a) 2ndmode shape, (b) 3rdmode shape, (c) 4th mode shape Theanalysis of graphspresented inFig. 8 can identify locationswheredamageoccurred.Very good resultswere obtained for the secondand fourthmode shapes.A comparison ofFigs. 8 and 7 reveals that changes determined with the experimental data are smaller than those obtained in Experimental validation of the use of energy transfer ratio... 251 numerical simulation. As mentioned above, this is due to the influence of external factors and measurement noise. More comprehensivedataon the experimental research, themodel and thenumerical analysis can be found elsewhere (Jarosińska, 2014). 5. Conclusions The paper presents results of sensitivity analysis conducted for natural frequencies, damping ratio and energy transfer ratio following damage introduced into the connection of a steel- concrete composite beam. Both numerical analysis results and experimental research confirmed that ETR is more sensitive to the beam damage than the vibration frequency and damping ratio. Interestingly, it is possible to effectively localise the damage introduced into the beam, based on the analysis of changes of the locally determined ETR. It must be stressed, however, that the determination of ETR requires experimental measurements to be conducted with high precision since the method is based on changes in composite mode shapes. Diagnostics of structures is amultistage process. Firstly, the damagemust be detected using modal parameters determined globally for thewhole structure: vibration frequency,modal dam- ping coefficient, global ETR, etc. Secondly, the detected damage is localised using locally deter- mined parameters, e.g. PMAC (Heylen and Janter, 1990), CDF (Wahab and deRoeck, 1999) or local ETRwhich has been used in the paper. Effectiveness of anymethodmay be dependent on external factors and measurement apparatus. The best way to detect and precisely localise the damage is to use several methods at the same time. Evaluation of changes inmodal parameters, including global and local ETR, when accompanied by other methods, may be a successful tool for detecting and localising damage in steel-concrete composite structures. The problem of damage detection and localisation in composite beams is going to be conti- nued in further research. Future investigations will be extended to include other types of damage and localisation techniques. References 1. AllemangR.J.,BrownD.L., 1982,Correlationcoefficient formodal vector analysis,Proceedings of the 1st International Modal Analysis Conference, Orlando, Florida, 110-116 2. DilenaM.,MorassiA., 2003a,Adamageanalysis of steel-concretecompositebeamsviadynamic methods: Part I. Experimental results, Journal of Vibration and Control, 9, 507-527 3. DilenaM.,MorassiA., 2003b,Adamageanalysis of steel-concretecompositebeamsviadynamic methods: Part II. Analytical models and damage detection, Journal of Vibration and Control, 9, 529-565 4. Dilena M., Morassi A., 2004, Experimental modal analysis of steel concrete composite beams with partially damaged connection, Journal of Vibration and Control, 10, 897-913 5. Dilena M., Morassi A., 2009, Vibrations of steel-concrete composite beams with partially de- graded connection and application to damage detection,Journal of Sound andVibration,320, 1/2, 101-124 6. Heylen W., Janter T., 1990, Extensions of themodal assurance criterion, Journal of Vibration and Acoustics, 112, 468-472 7. JarosińskaM., 2014,Detection of steel-concrete compoundbeams throughmodal analysis (inPo- lish), Doctoral disertation, ZachodniopomorskiUniwersytet Technologicznyw Szczecinie,Wydział Budownictwa i Architektury, Szczecin 252 T.Wróblewski et al. 8. Kong F., Liang Z., Lee G.C., 1996, Bridge damage identification through ambient vibration signature, Proceedings of the 14th International Modal Analysis Conference, Dearborn, Michigan, 717-724 9. Kruszewski J., Gawroński W., Wittbrodt E., Najbar F., Grabowski S., 1975, Rigid Finite Element Method (in Polish), Arkady,Warszawa 10. Kruszewski J., Sawiak S., Wittbrodt E., 1999,CAD CAM: Rigid Finite Element Method in Structural Dynamics (in Polish),WydawnictwaNaukowo-Techniczne,Warszawa 11. Lee G.C., Liang Z., 1999, Development of a bridge monitoring system, Proceedings of the 2nd internationalWorkshop on StructuralHealthMonitoring, StandfordUniversity,California, 349-358 12. Li J., Hao H., Xia Y., Zhu H., 2014, Damage detection of shear connectors in bridge struc- tures with transmissibility in frequency domain, International Journal of Structural Stability and Dynamics, 14, 2 13. Li J., Hao H., Zhu H., 2014, Dynamic assessment of shear connectors in composite bridges with ambient vibrationmeasurements,Advances in Structural Engineering, 17, 5, 617-638 14. Liang Z., LeeG.C., 1991,Damping of structures: Part 1 –Theory of complex damping,National Center for EarthquakeEngineeringResearch, Technical Report NCEER-91-0004, State University of NewYork at Buffalo, NewYork 15. Liang Z., Huang T.J., Lee G.C., 1997, A study on damage dynamics by using the modal energy transfer,Proceedings of the 15th InternationalModal Analysis Conference, Orlando,Florida, 1339-1345 16. Liu K., De Roeck G., 2009, Damage detection of shear connectors in composite bridges, Struc- tural Health Monitoring, 8, 5, 345-356 17. Rao S.S., 2004, Mechanical Vibrations, Fourth Edition, Pearsons Education International, New Jersey 18. Uhl T., 1997,CAD CAM: Computer Aided identification of Models of Mechanical Structures (in Polish),WydawnictwaNaukowo-Techniczne,Warszawa 19. Wahab M.A., De Roeck G., 1999, Damage detection in bridges usingmodal curvatures: Appli- cation to a real damage scenario, Journal of Sound and Vibration, 226, 2, 217-235 20. Wang T.L., Zong Z., 2002, Improvement of evaluation method for existing highway bridges, Technical Report FL/DOT/RMC/6672-818, Florida Department of Transportation, USA 21. Wang T.L., Zong Z.H., 2003, Structural damage identification by using energy transfer ratio, Proceedings of the 21st International Modal Analysis Conference: A Conference on Structural Dy- namics, Kissimmee, Florida 22. Wittbrodt E., Adamiec-Wójcik I., Wojciech S., 2006,Dynamics of Flexible Multibody Sys- tems. Rigid Finite Element Method, Springer-Verlag, Berlin, Heidelberg, NewYork 23. Wróblewski T., Jarosińska M., Berczyński S., 2011, Application of ETR for diagnosis of damage in steel-concrete composite beams, Journal of Theoretical and Applied Mechanics, 49, 1, 51-70 24. WróblewskiT., JarosińskaM.,Berczyński S., 2013,Damage location in steel-concrete com- posite beams using Energy Transfer Ratio (ETR), Journal of Theoretical and Applied Mechanics, 51, 1, 91-103 Manuscript received May 21, 2015; accepted for print July 17, 2016