5. Brückenkolloquium - September 2022 223 Multi-Sensor measurements on a large-scale bridge model Dr. Chun-Man Liao Bundesanstalt für Materialforschung und -prüfung (BAM), Berlin, Germany Dr. Konstantin Hicke Bundesanstalt für Materialforschung und -prüfung (BAM), Berlin, Germany Dr. Felix Bernauer Ludwig-Maximilians-University, Munich, Germany Dr. Heiner Igel Ludwig-Maximilians-University, Munich, Germany Dr. Celine Hadziioannou Institute of Geophysics, University of Hamburg, Hamburg, Germany Dr. Ernst Niederleithinger Bundesanstalt für Materialforschung und -prüfung (BAM), Berlin, Germany Abstract This contribution introduces an investigation of a large-scale prestressed concrete bridge model (“BLEIB” structure at the BAM-TTS open air test site) by means of on-site cooperative measurements. This bridge has an external post-tensioning system and has been instrumented with the ultrasonic transducers, temperature sensors and optical fibers for Distributed Acoustic Sensing (DAS). Our experiment was designed to test the suitability of the novel 6C sensors developed within the framework of the GIOTTO project - the IMU50. The IMU50 sensor enables vibration measurements in translation along three axes and rotation around three axes. The geophone sensors were considered for complementary measurements of vertical velocity response. In the experiment, several perturbations were achieved by controlling the external influence factors such as loading and prestressing changes. The aim of the integrated measurement strategy was to fully observe the results of the condition change and to verify the effectiveness of multiple sensors for bridge monitoring. 1. Introduction Recently, the topic of bridge monitoring has become a focus in civil engineering. A proper inspection of the bridge condition benefits reliable diagnosis of the structural performance and prediction of the service time of the structure. This is a way of economic management to achieve the safety and sustainability of the civil structures during their operation. Thus, the ability of structural health monitoring (SHM) in civil engineering is important. The damage identification based upon changes in vibration characteristics can be an approach that monitor changes in the structures on a global basis [1]. With the aid of advanced technologies in sensors, real time monitoring can be conveniently carried out. In the framework of the project Giotto (abbreviated from Gebäudeschwingungen: kombinierte Zustandsanalyse mit innovativem Sensorkonzept), the multi-sensor measurements on a prestressed concrete bridge model were carried out. The goal of the Giotto project is to develop real time monitoring methods for civil structures during normal operation and to advance the current damage detection methods. In this project, an innovative 6C prototype sensor IMU50 that was produced for the navigation by 6-components (6C) was designed. To validate the application of this sensor for the purpose of real time monitoring, vibration tests were undertaken while the reference sensors, which are broadband seismometers, were applied. The tilt correction of the translation acceleration output will be compared with the direct measurement of rotation. This contribution presents an investigation of a large-scale prestressed concrete bridge model (“BLEIB” structure at the BAM-TTS open air test site). The specific feature of this bridge is its adjustable prestressing force system and the mounted instruments, including the temperature sensors, the ultrasonic transducers as well 224 5. Brückenkolloquium - September 2022 Multi-Sensor measurements on a large-scale bridge model as the distributed acoustic sensing (DAS), which can make available spatially resolved dynamic strain information from which vibrational states under operational conditions can be determined [2]. The aim of performing on-site cooperative measurements is to fully observe the integrated influence on the bridge caused by the environmental and structural condition change. At the same time, not only the effectiveness of the multiple sensors for bridge monitoring can be proved, but also the validation of the innovative 6C prototype sensor IMU50 can be achieved. In the experimental analysis, the structural condition change regarding the altered prestressing force and the various loads, was identified by the modal frequencies. The temperature effects on the long-term monitoring of the ultrasonic waves and the vibrational waves were compared by the coda wave interferometry (CWI) method. The coincidence of the translation and the rotation responses was also proved. In addition, the advantage of the DAS measurement was revealed by identifying the modal shapes. This paper shows the preliminary study of the applicability of the multi-sensors for bridge structural health monitoring (SHM). The further study of damage detection should be done by the profound analysis of the obtained measurement data. 2. Experiments 2.1 Test structure The test structure is a large-scale prestressed concrete bridge model “BLEIB” (fig. 1), at the BAM-TTS open air test site in Brandenburg. This bridge has 24-meter length and 0.9-meter width. It contains a built-in unbonded post-tensioning system. Thus, the prestressing force can be precisely adjusted. Fig. 1: Large-scale bridge model „BLEIB“ with the post-prestressing system. 2.2 Sensors 2.2.1 Compact 6C prototype sensor IMU50 is a type of 6-degree-of-freedon (6D) Fibre Optic Gyroscopes (FOGs) sensors (fig. 2). This sensor uses three perpendicular silicon based capacitive Micro-Electo Mechanical System (MEMS) accelerometers to measure translational motion, and three FOGs coiled around the translational axes to measure rotations [3]. The combination of 3D rotational and 3D translational vibration responses can provide condition information without the need for a reference frame. Fig. 2: Left: IMU50 sensor. Right: IMU50 in water protector - pot. 2.2.2 Seismic vibration sensors To validate the innovative 6C prototype sensor IMU50 for SHM, the on-site cooperative measurements provide a comparative examination. The multiple sensors which can record structural vibrations as the reference for the IMU50 measurement results, are shown in fig. 3. The broadband seismometer - Trillium Compact was used for measuring the translational motions in the direction of three-axes. The broadband rotation sensor - blueSies-3A was used for measuring the rotation motions along the three axes. The geophones were applied for measuring the vertical vibration velocity responses of the bridge. These sensors can be placed on the top of the bridge for measurements. Fig. 3: Vibration sensors (from left to right): Trillium Compact, blueSeis-3A, geophone. 5. Brückenkolloquium - September 2022 225 Multi-Sensor measurements on a large-scale bridge model 2.2.3 Embedded sensors For the non-destructive testing (NDT), the piezoelectric ultrasonic transducers (fig. 4 left) - designed for embedding into concrete have been mounted as the “BLEIB” bridge was constructed. This transducer can be applied as the active source to send the ultrasonic waves, it also can be applied as the receiver. The ultrasonic wave was generated with a central frequency of 60 kHz. The sampling rate was 2MHz. During the measurements, 10000 samples were taken. To observe the scattering of the ultrasonic waves in the area where the direct path between source and receiver, the coda wave interferometry (CWI) technique was considered to extract the wave velocity variation. In addition, the dynamic strain rate corresponding to vibrations induced by excitation of the structure was measured along the optical fibers (fig. 4 middle) embedded close to the surface of and in axial direction of the “BLEIB” bridge using a commercial distributed acoustic sensing (DAS) device. The measurement approach is based on measuring spatially resolved Rayleigh optical backscatter signals following sequential optical interrogator pulses propagating through the sensing fiber. Local dynamic strain of the fiber (exerted by the structure’s vibrations) within the gauge length defined by the interrogator pulse duration results in a linear change to the backscatter signal’s phase from which the strain (rate) and thus local vibrations can be determined [4]. The pulse propagation results in effective continuous multiplexing and thus makes spatially resolved vibration information available. This therefore enables DAS along the structure. Fig. 4 (from left to right): Acsys SO807 [5], optical fibers glued into grooves on the bridge surface, TEWA sensor [6]. For the evaluation of the influence of environmental aspects on the bridge, temperature sensors (TT0210KC3- T105-1500 from TEWA sensors as shown in fig. 4 right) were applied. The temperature sensors are calibrated for a temperature range of −10 ◦C to 35◦C and provide precise measurements in this range [6]. The temperature was taken every 30 minutes to monitor the ambient temperature and the temperature change in the concrete. 2.3 Set up 2.3.1 Vibration sensors The sensor instrumentation can reference to fig. 5. From no. 1 to no. 12 are the positions of the geophones. The sampling rate was set 500 Hz. The distance between geophones was 2 m except that the distance between no. 6 and no. 7 was 1.6 m. From no. 13 to no. 16 are the positions of the 6D stations. At each station, a broadband seismometer - Trillium Compact and a broadband rotation sensor - blueSies-3A were applied. Their sampling rates were set to 200 Hz. The four IMU50 sensors were also placed next to the geophones at the positions no. 3, no. 6, no. 9 and no. 12. This can be seen in fig. 2 right. The sampling rate was set to 100 Hz. 2.3.2 Embedded sensors In fig. 5, the cross sections A, B, C, D and E are the locations where the ultrasonic transducers are embedded. Two transducers were mounted at each cross section, one at the east side and the other at the west side (fig. 6). The wave propagation between A and B, and between D and E were considered to evaluate the structural properties at the mid-spans, where the most cracks have progressed. Fig. 5: Sensor plan. 226 5. Brückenkolloquium - September 2022 Multi-Sensor measurements on a large-scale bridge model Fig. 6: Two ultrasonic sensors at a cross-section. 2.4 Vibration tests The ambient vibration measurements were performed two times, 1 day in 2020 and 2-3 weeks in 2021. Only the environmental factors without any structural condition change interfered with the vibration responses of the bridge. Two days in Oct. 2021 were chosen for the measurements as the “BLEIB” bridge was subjected to different vertical excitations under various external conditions, which were caused by the prestressing force change and the loading cases. Thereby, the natural damage was induced by opening the already existing cracks. Reference to Tab. 1, the prestressing force was rearranged. Firstly, the prestressing force was reset to 0 kN, then raised up to level 1 (450 kN). Afterwards, the prestressing force was decreased by the amount of 50 kN for every level until level 6 (200 kN). Thereafter, the prestressing force was enhanced by the amount of 50 kN for every level until level 11 (450 kN). Between level 2 and level 5, and at level 10, the loads 300 kg. 600 kg and 900 kg were applied to the bridge (position reference to fig. 5) to obtain the different event extensions of the already existing cracks. Tab. 1: The alteration of prestressing force and loads Reset 0 kN Load case Level 1 450 kN 0 kg Level 2 400 kN 0 kg 300 kg 600 kg 900 kg Level 3 350 kN 0 kg 300 kg 600 kg 900 kg Level 4 300 kN 0 kg 300 kg 600 kg 900 kg Level 5 250 kN 0 kg 300 kg 600 kg 900 kg Level 6 200 kN 0 kg Level 7 250 kN 0 kg Level 8 300 kN 0 kg Level 9 350 kN 0 kg Level 10 400 kN 0 kg 300 kg 600 kg 900 kg Level 11 450 kN 0 kg At each changed structural condition, the 10-minute ambient vibrations and the 3-minute impulse vibrations were recorded. 3. Environmental influences During the ambient vibration monitoring, no excitation was applied and the background noise at the “BLEIB” bridge was not caused by traffic vibrations but by wind. This section introduced the environmental factor which can influence the propagation velocity of ultrasonic waves in the bridge and the velocity of vibrational waves. 3.1 uncertainty of ultrasonics Temperature is known to have an influence on the ultrasonic pulse velocity of solid concrete [7]. Since the field conditions are not as pure as in the laboratory, Niklas et al. carried out a long-term velocity monitoring experiment outside the laboratory and tried to find a correlation between changes in temperature and ultrasound [8]. In recent years, coda wave interferometry (CWI), originally developed in seismology, has been applied to ultrasonic measurements in concrete [9]. In CWI application, a reference ultrasonic signal should be chosen for comparison of changes in signals during the monitoring. The change will be reflected on the velocity variation of the waves. A stretching factor is defined as equivalent to an average wave velocity variation, in eq. [1]. [1] where α is the stretching factor, t is time, v is velocity. By stretching the compared wave, its cross-correlation with the reference wave can help us to find the velocity change. The following eq. [2] is for calculating the crosscorrelation of the stretched wave and the reference wave. [2] where u c (t + αt) and u r (t) represent the stretched signal and the reference signal, respectively. Through eq. [2], the highest CC determines the best stretching factor. Considering the linear regression, the ratios of velocity change to the temperature 1 K were obtained, as shown in fig. 7. This indicates the uncertainty of ultrasonics due to temperature alteration in different months. 5. Brückenkolloquium - September 2022 227 Multi-Sensor measurements on a large-scale bridge model Fig. 7: Linear regression of velocity to concrete temperature between position A and B (position reference to fig. 5). 3.2 uncertainty of seismic waves The natural frequency can be obtained by means of Frequency Domain Decomposition (FDD) [10] technique, which is commonly used as algorithm for the operational modal analysis (OMA) [11] of civil structures. According to the ambient vibration recordings for 17 hours at the four 6D stations, the first modal frequency was changed from 3.62 Hz to 3.7 Hz. The alteration was not obvious. However, a reference was chosen to calculate the frequency variation as shown in fig. 8. Fig. 8: Difference of first mode frequency in the 17-hour monitoring. To calculate the velocity change of the vibrational waves during the 17 hours monitoring, the vibration signals were filtered in the range of 1-4 Hz. A cross-correlation function as the seismic interferometry technique was adopted to reconstruct the virtual wave propagation between two 6D stations. Fig. 9 was obtained by applying CWI to the seismic interferometry between two 6D stations. In fig. 9, the positions of station 1, 2, 3, 4 refer to no. 13 no. 16 in fig. 5. Fig. 9: Velocity variation of the vibrational waves filtered in the frequency range 1-4 Hz. The scale of the alternation in both fig. 8 and fig. 9 infer that the velocity change has a linear relation to the natural frequency. In addition, the tendency of the velocity change and the frequency change indicated the temperature change. The higher temperature, the lower natural frequency and lower wave velocity. And vice versa. On the other hand, the accuracy of the translation measurement and that of the rotation measurement is approved by their coincident results in fig. 8 and fig. 9. 4. Vibration data and dynamic strain data The DAS measurement provides truly spatially continuous vibration information by dynamic strain data. In contrast, the 6D station combining 3-component (3C) translation seismometer and 3C rotation seismometer as well as the IMU50 sensors can measure only the point vibration responses of the structure, albeit providing higher precision and accuracy. Obtaining modal parameters from the recordings can help to describe the vibration characteristics of the “BLEIB” structure. Thus, spectral analyses of the dynamic strain data and 6D vibrations were conducted. 4.1 Spectral analysis Power spectra for each measurement position along the sensing fiber were calculated via FFT of the distributed vibration information obtained from the commercial DAS device for no load and the prestressing force 400kN shown in fig. 10. Vibrations were induced by hammer blow simulating a pulse-like (broadband) excitation of the structure. A 20 s time window after the onset of the pulse was used to calculate the spectra. The position of the hammer blow can reference to fig. 5. 228 5. Brückenkolloquium - September 2022 Multi-Sensor measurements on a large-scale bridge model Fig. 10: Spatially resolved spectra of hammer blow-induced vibrations measured by DAS, with prestress force 400kN and 0 kg load. Fig. 10 shows a distributed view of calculated spectra along one fiber run on top of the bridge (24 m). Several low-frequency vibrational modes can be seen. The spatial distribution corresponds to typical harmonic mode shapes with increasing order for a vibrating structure. Fig. 11 depicts the onset of the the structures’s vibrations in a distributed way via the measured strain rate in a waterfall plot. Fig. 11: Onset of vibrations induced by hammer blow measured by DAS along bridge length. On the other hand, the impulse vibration responses subjected to the same vertical excitation recorded at a 6D station (referring no. 14 in fig. 5) is shown in fig. 12. Fig. 12: Impulse responses at a 6D station (no.14 in Fig. 5) under prestressing force 400kN and 0 kg load. Fig. 13: Vibrations measured as strain rate using DAS at measurement position corresponding to a 6D station (no.14 in Fig. 5) under prestressing force 400kN and 0 kg load. For comparison, fig. 13 shows a time trace of the dynamic strain rate measured by the DAS device at an individual measurement position corresponding to the location of 6D station no.14. The corresponding time-dependent spectrum of the impulse response as measured at the 6D station (fig. 12) is shown in fig. 14. Analogously, the spectrogram calculated from the fiber sensors strain rate measurement at the same location (fig. 13) is depicted in fig. 15. Fig. 14: Spectrum of translation responses at a 6D station (no.14 in fig. 5) under prestressing force 400kN and 0 kg load. Fig. 15: Spectrogram calculated from DAS strain rate measurement at location no.14 in fig. 5 following a hammer blow under prestressing force 400kN and 0 kg load. Comparing fig. 14 with fig. 15, the modal frequencies observed from the DAS spectrum and from the translation vibration spectrum show a good match, though the observed temporal persistence (damping) of the various modes differ, resulting among other reasons from the different mechanical couplings of the sensors. 5. Brückenkolloquium - September 2022 229 Multi-Sensor measurements on a large-scale bridge model Further, the spectra of the translation responses under the same condition at four 6D stations were compared as shown in fig. 16. The distinct peaks indicate the first three modal frequencies 3.6 Hz, 5.8Hz and 15Hz. the positions of station 1, 2, 3, 4 refer to no. 13 no. 16 in fig. 5. Fig. 16: Spectra of translation responses at four 6D stations under prestressing force 400kN and 0 kg load. 4.2 Modal frequency change Furthermore, the modal frequencies were determined from the DAS data and from the 6D station measurements via OMA method, respectively. The comparison of the first 3 modal frequencies in fig. 17 shows a very good agreement and the same behavior under different conditions: the higher the prestressing force, the higher the mode frequencies and the heavier the loading, the lower the resulting frequencies. Fig. 17: Left: modal frequencies from DAS measurement data. Right: Modal frequencies tend obtained from translation vibration data via OMA. 5. Conclusion and outlook The IMU50 sensors, geophones and 6D broadband seismometers, as well as ultrasonics and DAS techniques were applied to fulfill the on-site cooperative measurements. The preliminary investigation was conducted by the vibration assessment and the ultrasonic analysis. From the results of the 6D vibration monitoring, the consistence of translation and rotation measurements by use of different sensors was confirmed. In the modal analysis of the 6D vibration recordings and the DAS measurements, the changes in the prestressing force and the loads were identified. Nevertheless, the deviation of the modal frequencies in DAS and in translation vibration results need to be examined. The reason has not yet discussed in this contribution. In addition, the CWI method was applied to show the influence of the temperature on the ultrasonics in the “BLEIB” bridge. The temperature effect was quantified by the linear regression of the velocity variation with respect to the concrete temperature. This can be used to dismiss the uncertainty of the temperature in the ultrasonic measurements to identify the velocity change in dependence of the prestressing force and the loads. Afterall, the assessment of the IMU50 sensors was not presented in this study. To validate the IMU50 sensors in SHM is our next work. Moreover, we tend to draw a distinction from the local/ global damage consequences of bridges and the prestress loss in the further study of this comprehensive measurements. 6. Acknowledgement This research is part of the Giotto Project financially supported by Bundesministerium für Bildung und Forschung (BMBF) in the frame of “Früherkennung von Erdbeben und ihren Folgen” program (Grant No: 03G0885D). The authors thank the colleagues from BAM division 8.2 and 8.6 for their support of our experiment. Literature [1] Farrar, C. R., Doebling, S. W., & Nix, D. A.: Vibration-based structural damage identification. In: Philosophical Transactions of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences, 359(1778)/ 2001, pp. 131-149. [2] Hicke, K., Hussels, M., Eisermann, R, Chruscicki, S. and Krebber, K.: Condition monitoring of industrial infrastructures using distributed fibre optic acoustic sensors. In: 25th Optical Fiber Sensors Conference (OFS), IEEE/ 2017. [3] Murray-Bergquist, L., Bernauer, F., & Igel, H.: Characterization of six-degree-of-freedom sensors for building health monitoring. In: Sensors, 21(11)/ 2021, pp. 3732. [4] He, Z. and Liu, Q.: Optical Fiber Distributed Acoustic Sensors: A Review. In: J. Lightwave Technol. 39/ 2021, pp. 3671-3686. 230 5. Brückenkolloquium - September 2022 Multi-Sensor measurements on a large-scale bridge model [5] Niederleithinger, E., Wolf, J., Mielentz, F., Wiggenhauser, H., & Pirskawetz, S.: Embedded ultrasonic transducers for active and passive concrete monitoring. In: Sensors, 15(5)/ 2015, pp. 9756-9772. [6] Fontoura Barroso, D., Epple, N., & Niederleithinger, E.: A portable low-cost ultrasound measurement device for concrete monitoring. In: Inventions, 6(2)/ 2021, pp. 36. [7] Niederleithinger, E., & Wunderlich, C.: Influence of small temperature variations on the ultrasonic velocity in concrete. In: American Institute of Physics (AIP) Conference Proceedings, vol. 1511, no. 1/ 2013, pp. 390-397. [8] Epple, N., Barroso, D. F., & Niederleithinger, E.: Towards Monitoring of Concrete Structures with Embedded Ultrasound Sensors and Coda Waves- First Results of DFG for CoDA. In: European Workshop on Structural Health Monitoring, Springer, Cham./ 2020, pp. 266-275. [9] Planès, T., & Larose, E.: A review of ultrasonic Coda Wave Interferometry in concrete. In: Cement and Concrete Research 53/ 2013, pp. 248-255. [10] R. Brincker, P. Andersen, and N. J. Jacobsen: Automated frequency domain decomposition for operational modal analysis. In: IMAC-XXIV: A conference & exposition on structural dynamics. (Society for Experimental Mechanics, Bethel, CT 2007). [11] B. Peeters, and G. De Roeck: Stochastic system identification for operational modal analysis: a review. In: J. Dyn. Sys., Meas., Control, 123(4)/ 2001, pp. 659-667.