crucial for results of the bridge monitoring based on vibration tests. .... This method of FRF identification assumes application of a broadband excitation .... train passage and free vibration after the train has passed the bridge â labelled âforced.
International Conference "Sustainable Bridges – Assessment for Future Traffic Demands and Longer Lives", Wrocław, Poland, October 10-11, 2007
Inertial exciter as a tool for dynamic assessment of railway bridges Jaroslaw ZWOLSKI, Józef KRZYŻANOWSKI, Paweł RAWA, Wacław SKOCZYŃSKI & Janusz SZYMKOWSKI Wroclaw University of Technology, Poland
ABSTRACT: Accurate and precise identification of modal parameters of railway bridges is crucial for results of the bridge monitoring based on vibration tests. In this paper a monitoring system consisting of exciting and measuring hardware and developed control software was described. The principles of modal parameters identification and methodology of bridge monitoring by repeating vibration tests are presented together with the system application to tests of a railway bridge. Comparison of tests results with use of exciter and freight trains passing the bridge as a sources of excitation confirmed efficiency of the monitoring system.
1. INTRODUCTION For structures along railway lines – and especially high-speed lines – all dynamic issues are very important. The base for theoretical and experimental dynamic analyses of structures is modal analysis and its results: modal frequencies, damping ratios and mode shapes. Modal models consisting of these quantities are extensively used in various fields: • damage detection – some types of damages (e.g. material losses, cracks, loosening of connections) influence stiffness of the bridge structure and consequently cause changes of the modal parameters – the changes of the modal parameters can be investigated by means of vibration tests; • model updating – on the basis of precise experimental data collected during vibration tests the methodology of structure modelling and assessment can be improved and applied in analysis of new built and existing structures; • sensitivity analysis – answer the question: what should be changed in structure (add a mass, a stiffener or a damper, modify boundary conditions) to change structure modal parameters and consequently to improve performance – decrease vibration amplitudes, cut down on noise etc. Successes of these analyses strongly depend on accurate and precise identified modal models of structure. For high accuracy of results of modal analysis independency of excitation characteristics is required. When an excitation technique commonly used for bridges is applied
(e.g. vehicles passing the bridge) the influence of the excitation characteristics (e.g. mass of the moving train, boogies suspension properties etc.) on vibrations parameters is habitually neglected. Precision of the results of modal estimation is often identified as resolution of spectra in frequency domain what directly depends on the data acquisition time. The second aspect of precision is repeatability of the results obtained in consecutively executed tests. In this paper technology of bridge vibration tests by means of inertial exciter is presented. Practical application of the technology is illustrated by comparison of modal parameters of a railway bridge identified with application of moving freight trains and using a rotational eccentric mass (REM) exciter. The comparison is focused on accuracy and precision of results achieved by application of the two excitation techniques and presents practical aspects of field tests of bridges. A research team from Wroclaw University of Technology (WUT) taking part in the project “Sustainable Bridges – Assessment for Future Traffic Demands and Longer Lives” worked out a method of railway bridges testing with use of a rotational mass exciter (SB5.6 Shaker, 2007). The testing system consisting of a measuring device with sensors, computer, exciter and controllable inverter works under control of software MANABRIS. Taking advantage of such characteristic of tests with exciter as repeatability, force control and measurement, precision and accuracy in structure’s modal parameters estimation – the testing system can be exploited not only for single tests of bridges (Bień and Zwolski, 2007) but also for bridge monitoring (SB8.2 Demo, 2007). 2. MODAL PARAMETERS IDENTIFICATION Identification of modal parameters characterizing a bridge structure can be done using two approaches: theoretical and experimental. The first method assumes creation of a theoretical (the most often numerical) model of the analyzed structure and execution of Theoretical Modal Analysis (TMA). Customarily it is carried out by means of computational software based on Finite Element Method (FEM). Results are obtained in form of eigenfrequencies and eigenmodes (mode shapes). The experimental approach relies on analysis of data measured during test when the structure’s vibrations are excited by various forces. The algorithm of data processing and analysis depends on the technique of excitation of the structure vibrations. When the excitation forces are measured during the experiment the method of data processing is called Experimental Modal Analysis (EMA) and the only practically usable source of excitation is an exciter. If other techniques of excitation are used: moving vehicles, normal traffic, wind, waves, microtremors etc. the acting forces are immeasurable and the method is called Operational Modal Analysis (OMA). In EMA the modal parameters are identified from Frequency Response Functions (FRF) determined using measurements of excitation force and structure’s responses in the following formula (Ewins, 1999):
H 1 (ω ) =
G XF (ω ) G FF (ω )
(1)
where: H1 (ω ) – estimator of the FRF, G XF (ω ) – cross-spectrum of the measured displacement, velocity or accelerations with the measured excitation force, GFF (ω ) – autospectrum of the measured excitation force, ω – frequency. For FRF estimation the spectra G XF (ω ) and GFF (ω ) are averaged taking into account results of few experiments repetitions.
b)
a)
c)
d)
Figure 1. Various exciters for bridge testing: a) rotational eccentric mass (REM) exciter – Zwolski (2007), b) vertical, driven by electrohydraulic actuator – Krämer, De Smet and Peeters (1999) c) horizontal, driven by electrohydraulic actuator – Ye, Fanjiang and Yanev (2005), d) sequence impulse exciter – Bień et al. (2004)
Exciters are mechanical devices generating forces of various types: harmonic (sinusoidal), sweep (harmonic with tuned frequency), random etc. and enabling measurement of the force during experiments. The most known devices from literature are driven by electric motor, hydraulic or pneumatic actuator or electromagnet. Some examples of exciters are presented in Figure 1. This method of FRF identification assumes application of a broadband excitation technique: in theory white noise excitation and in practice Gaussian random, pseudo random, sweep or impulse excitation. In the case of application of harmonic excitation the FRF values calculated using formula (1) are valid for the excitation frequency only. For FRF identification within selected frequency range the harmonic excitation is executed at predefined values of frequency and discrete values of FRF are calculated for each excitation frequency. The procedure is called Stepped Sine Test (SST) and usually is perceived as a time consuming but very precise method. A supplementary technique is used for mode shape identification (MSI) – at steady-state resonance excitation (with the determined resonance frequency) sensors are moved from point to point. The amplitude and phase of the acquired signals in proportion with the reference signal from one or more fixed sensor enables identification of the investigated mode shape.
On the basis of FRF identified for all structure degrees of freedom (DOF) all modal parameters are determined using one or more methods of estimation. Many of them are based on curve-fit algorithms applied to the FRF directly in frequency domain or after its transformation to time domain by Inverse Fourier Transform (IFT). The most effective methods of modal estimation are presented by Ewins (1999) as well as by Maia and Silva (1997). 3. MONITORING SYSTEM The idea of monitoring of the bridge condition based on observation and tracking of its modal properties, known as Structural Health Monitoring (SHM), was described e.g. by Brownjohn et al. (2004). Similar premises were a genesis of the railway bridge monitoring system which enables assessment of structure condition based on results of repeated testing sessions. The monitoring system developed by the research team from WUT consists of (Figure 2): • REM exciter, programmable inverter & force sensors, • portable computer, • control software MANABRIS – see Figure 3, • measuring device & response sensors, • portable power generators, • measuring device & temperature and humidity sensors.
Figure 2. Elements of the monitoring system (SB5.6 Shaker, 2007)
When a REM exciter is used identification of modal parameters of the monitored structure can be performed in two ways: • by execution of the tests sequence employing the harmonic signals stepped over a range of frequencies – SST&MSI together with so known damping estimation techniques as Half Power Bandwidth Method or Logarithmic Decrement Method, • by execution of the sweep test repeated several times and averaging the results. Values of resonance frequencies and mode shapes should be compared with the values obtained in theoretical modal analysis and should be used for calibration of the model of structure. Sets of results, calculated and measured, are both stored in a database of a dedicated computer-based system. In this way the results of initial testing session create bases for monitoring process. Each next testing session is carried out in the same manner and with the same testing setup – the test parameters are taken from the structure’s database (Figure 4).
a)
b)
c)
Figure 3. Software MANABRIS – graphic interface during excitation and measurement: a) chart with the measured values, b) list of monitored sensors with currently acquired value and signal stationarity control, c) list of the scheduled excitations in SST (step 0.1 Hz, 24000 samples each excitation), graphical control of the excited frequency and progress bars of test (Zwolski, 2007)
Comparison of the structure modal parameters obtained during current test session with the results from the previous sessions enables concluding about technical condition of the monitored structure. The lack of any significant changes of the compared parameters means that there is no changes of the construction technical condition and a date of the next monitoring session can be planned (if monitoring is still needed). Considerable differences of compared modal parameters mean that the structure technical condition has changed. In such a situation the following levels of condition assessment can be applied: • elementary level – changes of modal parameters are evaluated on the basis of the first few vibration modes to make a decision if the structure requires additional detailed inspections or tests because its condition is reduced; • advanced level – changes of modal parameters are investigated on the basis of all available vibration modes what can enable identification of damage or structural modification type as well as damage parameters (location, intensity etc.); this information is used for a more precise evaluation of structure condition. Analysis on the advanced level gives more valuable results but it is of course more timeand cost-consuming than the elementary-level analysis. The testing system and all incorporated procedures of test execution and data processing came under detailed examination focused on efficiency, accuracy and precision of the method and applied equipment (Bień et al., 2006). One of those tests is presented in chapter 4.
Figure 4. Procedure of bridge condition monitoring based on the vibration tests (SB5.6 Shaker, 2007)
4. EXAMPLE OF VIBRATION TEST – OBJECTIVES AND SETUP The tests objectives were assumed as follows: • calibration of the measuring parameters (sampling frequency, period of data acquisition, etc.), • test of the control software MANABRIS focused on its stability and usefulness in field tests, • check of efficiency of all procedures implemented in the software used for test execution and preliminary data processing, • evaluation of the influence of real field conditions (traffic in the neighbourhood, operation in strong electric field etc.) on the system efficiency, • estimation of time required for vibration tests (time of train traffic disruption), • comparison of tests results with application of train passage and exciter as excitation sources.
a)
b)
c)
Figure 5. Bridge over Ślęza River in Wrocław during tests: a) general view from upstream side, b) view along the track with the exciter on the top, c) sketch of sensors location
For the system performance tests a steel simple supported bridge was chosen – Figure 5. The bridge is located in Wrocław over Ślęza River and carries one track on the ballasted deck made of orthotropic steel plate. The span length is 31.0 m and the structure is skewed. Program of the tests consisted of: • test with sweep excitation tuned exponentially in range 3-24 Hz, acquisition time 212 sec., sampling frequency 800 Hz – 3 repetitions, • test with sweep excitation tuned linearly in range 3-24 Hz, acquisition time 316.75 sec., sampling frequency 800 Hz – 2 repetitions, • SST with harmonic excitation in range 3-13.2 Hz with step from 0.016 to 0.032 Hz and with variable acquisition time from 31 to 62 sec., • test with passage of freight train – 5 repetitions. In the tested system a REM exciter is applied as a source of excitation (Figure 1a). To collect response of the structure and a set of inductive accelerometers and LVDT’s was used (Figure 5c) together with data acquisition system Spider8 manufactured by Hottinger Baldwin Messtechnik. Time of data acquisition in SST was 2 hours, in sweep tests 11 minutes (the same for both types of tuning) and for the tests with train passages – total acquisition time was 5 minutes. Time for installation, equipment check out and bridge tests with the exciter was short due to tight time between scheduled trains. Whole testing session took around 6 hours and 5 persons took part in the work. Uncomplicated installation of the exciter on the rails was possible thank to special wheels enabling easy transportation as well as simple crank device which aided location of the exciter on the supporting frame in the testing position. Results of all tests were processed in software MANABRIS. For all sweep tests FRF’s were determined by averaging results of repetitions and for SST values of response functions were
calculated for 121 frequency steps in the predefined range – Figure 6. In the case of tests with trains two separate autospectra were calculated: • for whole signal record, neglecting the difference between forced vibration during the train passage and free vibration after the train has passed the bridge – labelled “forced vibrations” in Figure 6, • for the sections of signal acquired after the train passage – labelled “free vibrations” in Figure 6. The obtained modal parameters are presented in Table 1. Their estimation regarding the tests with exciter was done using methods: Peak Picking (PP), Circle Fit (CF) and Line Fit (LF) – for details of the methods see Ewins (1999) and Maia and Silva (1997). Vibration frequencies identified in the tests with trains was found by means of PP method. The comparison of the obtained FRF’s and autospectra lead to the following remarks: • General shape of all determined FRF’s is consistent; in the range 3 – 15 Hz using PP method 5 modes were identified with frequencies: 4.029 Hz, 4.674 Hz, 8.956 Hz, 11.627 Hz and 14.525 Hz (average values for all estimation methods of modal parameters are given in Table 1). • Technique of harmonic excitation enables determination of FRF with the highest coherence and the lowest noise. FRF’s determined in the tests with both sweep excitations have the finest resolution (0.0047 Hz at the exponential sweep and 0.0032 Hz at the linear sweep) but in the range of low frequencies (below 4 Hz) and between resonances there is noise observed of remarkable intensity. It is caused probably by low excitation energy in low frequency range what is characteristic of REM exciters. Increasing the number of repetitions and averages as well as increasing the time of sweep excitations would be an effective means to cut down on intensity of noise in the FRF between resonances. 0.25
1.00 exponential sweep linear sweep harmonic forced vibrations free vibrations
2
2
abs(Ha) [m/Ns ]
4.6 Hz
0.80 4
4.677 Hz
0.20
14.557 Hz
Autospectrum [m /s ]
4.291 Hz
0.15
0.60
0.10
0.40
4.030 Hz
11.613 Hz
0.05
14.6 Hz
0.20
9.030 Hz 0.00 2.6
3.6
4.6
5.6
6.6
7.6
8.6 9.6 10.6 Frequency [Hz]
11.6
12.6
13.6
14.6
0.00 15.6
Figure 6. Comparison of FRF’s determined in tests with exciters and autospectra obtained in tests with trains passages – results for A02 sensor (refer to Figure 5c)
Table 1. Comparison of modal parameters obtained in the tests with train passages (DAE) and with the exciter (DAM) – identification using methods PP, CF and LF
Mode No.
DAE test Forced Free vibrations vibrations PP fr [Hz]
DAM test PP fr [Hz]
fr [Hz]
CF
LF
ζr
fr [Hz]
ζr
1
–
–
4.030
–
[%] –
4.027
[%] 1.1%
2
4.291
4.6
4.677
4.656
1.0%
4.689
1.0%
3
–
–
9.030
8.965
3.9%
8.874
5.0%
4
–
–
11.613
11.647
3.9%
11.622
3.8%
5
–
14.6
14.557
14.508
2.5%
14.511
2.1%
Passages of trains excited structure’s vibration in the range of low frequencies – the dominant frequency during the train passage is 4.291 Hz and it is lower than the frequency identified from accelerance by 0.386 Hz (9%). Free vibrations frequency for the first mode is equal to 4.6 Hz, what is close to the result of the tests with the exciter but the resolution of the autospectra is only 0.2 Hz. • In the frequency range between 6.6 Hz and 8.0 Hz in the autospectra some peaks are visible which are not confirmed by the tests with the exciter. The result can be interpreted as the effect of forced vibrations induced by boogies’ wheels passing the rail joints. The presence of the peaks makes the interpretation of the results as well as modal parameters identification more difficult. •
5. CONCLUSIONS Results of all preliminary calibrations and examinations of the monitoring system carried out in frame of the research done by WUT taking into account results of the presented test can be commented as follows: • The results of the tests showed that application of trains passages as a source of excitation in modal tests of this bridge is inefficient due to mass of the train decreasing natural frequency of the span during the passage and due to quite high damping values of the bridge and consequently very short time of free vibration after passage what implies low resolution of spectra. • Active vibrations excitation by means of exciter makes modal parameters identification less dependent on structural damping than free vibration tests. In the case of structures with high damping the acquisition time of valuable signal after the impulse excitation is short what implies low resolution of characteristics in frequency domain. • The main advantage of the testing method using exciters is that the accuracy of modal parameters identification is independent of the excitation source characteristics. Use of exciter enables recording the exciting force and structure response in time series of any length. Results of tests have shown that this excitation technique has such a properties as repeatability, precision, controllability of the excitation force and high signal-to-noise ratio what makes the technique a desirable tool in bridge monitoring based on vibration tests.
The developed software MANABRIS enables execution of all described tests in semiautomatic manner with precise control of all required parameters. Implemented graphic user interface and procedures for preliminary data processing are useful during field tests due to direct control of the measured data. • Proposed architecture of the testing system seems to be useful and comfortable to use. Integration of all tasks in one software application enables using one computer to control the inverter, the exciter and the measuring device. The portable power generator enables carrying out tests independently of local power sources. •
ACKNOWLEDGEMENTS This research is sponsored by EC within 6 Framework Project “Sustainable Bridges – Assessment for Future Traffic Demands and Longer Lives”. This support is greatly acknowledged. REFERENCES Bień J., Krzyżanowski J., Rawa P., Zwolski J. (2004): Dynamic Load Tests In Bridge Management, Archives of Civil and Mechanical Engineering, vol. 4, nr 2, p. 63–78. Bień J., Krzyżanowski J., Rawa P., Skoczyński W., Szymkowski J., Zwolski J. (2006): System for monitoring of steel railway bridges based on forced vibration tests, 3rd International Conference on Bridge Maintenance, Safety and Management, IABMAS, Porto, Portugal, July 16-19. SB5.6 Shaker (2007): Prototype of exciter for vibration tests and concept of monitoring system, Background document D5.6 to SB-Monitor (2007): Guideline for Monitoring of Railway Bridges. Prepared by Sustainable Bridges – a project within EU FP6. Available from: www.sustainablebridges.net [cited 30 November 2007]. Bień J., Zwolski J. (2007): Dynamic Tests in Bridge Monitoring – Systematics and Applications, 25th International Modal Analysis Conference, Orlando, Florida, USA. Brownjohn J., Tjin S.-C., Tan G.-H., Tan B.-L., Chakraboorty S. (2004): A Structural Health Monitoring Paradigm for Civil Infrastructure, 1st FIG International Symposium on Engineering Surveys for Construction Works and Structural Engineering, Nottingham, United Kingdom. Ewins D.J. (1999) Modal Testing: Theory, Practice and Application, Research Studies Press Ltd., Letchworth, Hertfordshire, UK (2nd Edition). SB8.2 Demo (2007): D8.2 Demonstration of bridge monitoring, Prepared by Sustainable Bridges – a project within EU FP6. Available from: www.sustainablebridges.net [cited 30 November 2007]. Krämer C., De Smet C.A.M., Peeters B. (1999): Comparison of ambient and forced vibration testing of civil engineering structures. Proceedings of 17th International Modal Analysis Conference, Kissimmee, FL, USA, p. 1030–1034. Maia N.M.M., Silva J.M.M. (1997): Theoretical and Experimental Modal Analysis, Research Studies Press Ltd, Hertfortshire. Ye Q., Fanjiang G.-N., Yanev B. (2005): Investigation of the Dynamic Properties of the Brooklyn Bridge, Sensing Issues in Civil Structural Health Monitoring, Farhad Ansari (ed.), Springer, Netherlands, p. 65– 72. Zwolski J. (2007): Identification of Bridge Structures’ Modal Parameters Applying Exciters. PhD Thesis 5/2007, Wrocław: Institute of Civil Engineering, Wroclaw University of Technology, 313 pp. [cited 31 September 2007].
