Dynamic tests of certain lattice girder railway bridge type nodes

Experimental Vibration Analysis for Civil Engineering Structures EVACES 09 Dynamic tests of certain lattice girder railway bridge type nodes S. Pradelok The Silesian University of Technology, Gliwice, Poland ABSTRACT: The paper describes results of dynamic tests of lattice girder railway bridge. The research project was to find out reasons behind lattice girders nodes cracking of railway bridge. The study outlines history of several previous investigations, that have not have convincingly explained reasons of failure. The lattice girders that are subject of the study are a part of a railway bridge, which consists of six independent, freely supported steel spans. Four of them are plate girders and two other are lattice girders. In one of lattice girders cracking was observed. In second, twin lattice girder, supporting the same track, such cracking did not occur. The research project focused on specific, particular circumstances for comparative investigation. Detailed models of four types of such nodes were analyzed. The study suggests that the main reasons of damage are local static and dynamic influences and real level strains.. INTRODUCTION Hard to explain failures occurred in a railway lattice girder bridge. Despite earlier conducted analysis and examinations the reasons behind these failures remained hard to rationalize. The structure, which was affected, is a part of a larger bridge consisting of six independent freely supported steel spans. They are made of girders four of them plate, two remaining lattice. No damages were observed in the plate girders, while one of the lattice girders displayed cracks in the bottom chord. Fissures were present in nodes next to the split of the concrete platform of the bridge. Multiple, detailed examinations did not reveal any cracks in the other, identical, lattice girder span, at the same railway track. That fact was later used to conduct comparison tests and analysis. In order to explain the reasons behind cracking four detailed calculation models of the girder nodes were built. First one of the very node that failed, second of the corresponding node in the twin girder, third and fourth of the nodes after retrofitting. As a result of a conducted analysis of the models it was proven, that the reason behind the failure were local static and dynamic influences acting on top of the overall stress level. Modern numerical methods allow to obtain very detailed results. Obviously the result quality is directly related to the precision and relevance of the input data. Therefore the results should be verified in empirical tests. Such tests were performed on the two lattice girders. The bridge was also subjected to dynamic testing. None of the test however attempted to test local vibration of the nodes. Test of that nature were performed now, and are described herein. According to the test program the sensors were located in a way that made possible measurements of local dynamic influences in the girders nodes. In this study only the result of retrofitted girder are presented.. TESTING EQUIPMENT Testing equipment was built in Roads and Bridges Department of The Silesian University of Technology. It consists of: 5

EVACES 09 Vibration analysis of bridges a portable computer, 6 PCMCIA measurment CARD DAQCARDAI6XE50, combining card, induction displacement sensors (P), acceleration sensors (A), 8 resistance tensometers (T). Figure. Measurment set Figure. Interactive panel The measurement card is a very versatile tool, and it had to be programmed for the purpose of this research. Purpose fit software was created for registration, monitoring and analysis of collected data National Instruments LabView graphical environment which allows use of ready made functions and procedures prepared as so called virtual instruments Figure shows interactive panel for anaylsis. It imitates as a real front plate of measurement equipment. The panel s dials and controls are operated with the computer keyboard and mouse. 5

Dynamic tests of certain lattice girder railway bridge type nodes 3. FIELD TESTS To determine dynamic characteristics of the bridge an identification experiment is conducted. The structure is subjected to a controlled excitement. It may be harmonic, with continuous alteration of frequency (sweep). Through a set of sensors the structure response is measured. Appropriate number of sensors needs to be placed to capture longitudinal, deflectional and torsional shapes of natural vibrations. Proper and effective placement of sensors requires deep knowledge and experience. Sometimes it is required to repeat the measurement and averaging of several series of measurements. The results are always related to frequencies (natural frequencies spectrum and associated shapes) and time, with full description of the testing process. Next step is analysis of the results. Received results are subjected to various numerical transformations. The results are only perceivably more precise, as they are only as good as the precision of the measurements, and the results should be always treated with a prudent caution. During the tests sensors reading were registered. In particular stress, displacement and acceleration and dynamic characteristic of selected nodes of the structure were recorded. induction sensors (one per girder) were installed to capture time changes of the girders deflections. Sensor P was installed in vicinity of node nr 4 (middle of the span), sensor P3 by node 3, /3 the girder span. Figure 3. Induction displacement sensors locations Accelerometers and tensometers were used to register local dynamic influences. Detailed survey was performed on node 5 of the left girder (looking toward Maczki). The sensors were installed to register locar vibrations in that particular node. Two accelometers A and A were placed in vertical alignment, on both sides of top flange of the plate girder, close to the joint of gusset plate of node 4. Accelerometer A was attached to the inside, and A to the outside of the span. 8 tensometers T T8 served to capture changes in the structure deflection. Those with odd numbers on the inside, the remaining ones on the outside. Four tensometers T, T, T7 and T8 were stick horizontally close to gusset plates, in the middle of top flange of the lattice girder thickness. T and T were placed on the node 4 side, T7 and T8 on the node 6 side. Further 4 tensometers T3 T6 were placed vertically on gusset plated at their edges. T3 and T4 on the node 4 side, T5 and T6 on the node 6 side. Dynamic load was simulated by a set of 3 electric railway engines ET4. It was attempted to simulate many real situation that the bridge would be subjected to in normal operation. Test programme consisted of runs at 0, 30, 50 and 70 km/h speeds (.9, 8.3, 3.9, 9.4 m/s) 53

EVACES 09 Vibration analysis of bridges Figure 4. Electro resistant tensometers (T) and akcelerometers (A) 54

Dynamic tests of certain lattice girder railway bridge type nodes 4. ANALYSIS OF REGISTERED DATA Of all results registered only selected, representative set is presented herein. They were collected for load movements in both directions (Maczki and Jęzor) on the lattice girders after retrofitting. These results form a good basis for assessment of quality of theoretical calculation models, which enables verification of results of modal and time spectrum analysis calculated results. Registered results of the tests are presented on the drawings below. Each drawing has 4 parts, a through d. Each parts has a heading, vibrogram, corresponding graph of power spectrum density (PSD) and a table with sensor id, amplitudes (/) and spread () and their changes in time t, type and order of the filter and identified, dominant frequencies of vibrations F max. 4.. Induction displacement sensors (P) Induction displacement sensors (P) were used to measure changes in time of deflection of main girders. Location of the displacement sensors enabled registration of vertical modes of the girders vibrations. Figure 5 and 6 shows exemplary, registered deflections over the time for 30 and 50 km/h load runs towards Maczki or Jęzor. Each drawing shows: Entire reregistered record of deflections without the signal post processing. It is impossible to visually asses the displacements. Only deflection spread u z () can be determined. As above, but post processed with Bessel band filter of 5th order, with lower frequency F d = Hz and top one F g = Hz. It is possible to determine dominant frequencies F max and spread of deflections u z () after deduction of quasi static signal. Part of registered measurement signal while the load is present on the bridge, after post processing with Bessel band filter of 5 th order, with lower frequency F d = Hz and top one F g = Hz.. It is possible to determine dominant frequencies F max and spread of deflections u z () after deduction of quasi static signal. Part of registered measurement signal after the load has left the bridge, after post processing with Bessel band filter of 5th order, with lower frequency F d = Hz and top one F g = Hz. It is possible to determine dominant frequencies F max and spread of deflections u z () after deduction of quasi static signal. The vibrograms shows how the vertical deflection changes during the passage of the load u z (t) on the main girder. Biggest spread of deflections u z ( ) reached 9,6 mm. It was registered by P sensor with load traveling at 70km/h toward Jęzor. After post processing of the signal from P sensor with Bessel band filter of 5th order, with lower frequency F d = Hz and top one F g = Hz (graphs b, c, local heap can be observed. Frequencies which correspond to the heaps were identified as F max. After filtering out all registered signal (part a very clear dominant heap of power spectrum density can be observed. Its frequency F max falls between,55 Hz do,8 Hz. Similar results is obtained from analysis of measurement signal while the load is on the bridge (part. Frequency of first F max falls then into range between,70 Hz and, Hz. After the load leaves the bridge however (part first maximum PSD occurs with frequency F max in range between,7 Hz and,37 Hz. Moreover the P sensor registered another heap with F max in range between 4,88 Hz and 8,73 Hz. Figure 6 shows exemplary results registered with sensor P3. All registered values are collected in table. Values shown on graphs are shown in corresponding columns:, in column and so on. Table. Results registered with sensors P and P3 Sensor u z () Identified F max [ Hz] P 8,94 9,6,55,8,70,,7,37 P3 3,66 4,58,78,35,80,0,00, In summary it can be stated that after filtering out parts of b, c, d values registered by P i P3 first heaps appeared at frequency F max, between,55 Hz and,37 Hz. In this range theoretically calculated frequency of second mode of natural vibration is located at,78 Hz. Frequencies F max for next registered heaps correspond to higher modes of natural vibrations. 5

EVACES 09 Vibration analysis of bridges P 6.30.76 8.99 Brak 0.03 P 0.435 0.504 0.938.55.70 P 0.435 0.504 0.938.70 P 0.83 0.97 0.580.7 7.43 Figure 5. Sensor P. Run at 50km/h. Toward Maczki J30a Jazda z prędkością 30 km/h w kierunku stacji Jęzor J30a Jazda z prędkością 30 km/h w kierunku stacji Jęzor P3 6.9 7.59 3.658 Brak 0.0 J30a Jazda z prędkością 30 km/h w kierunku stacji Jęzor P3 0.45 0.7 0.46.0.08.4 J30a Jazda z prędkością 30 km/h w kierunku stacji Jęzor P3 0.45 0.7 0.46.5 5.50 P3 0.9 0.89 0.48.00 Figure 6. Sensor P3. Run at 30km/h. Toward Jęzor 56

Dynamic tests of certain lattice girder railway bridge type nodes 4.. Accelerometers (A) Jazda z prędkością 50 km/h w kierunku stacji Maczki M50a 0.87 0.90.736 5.44 4.8 8.6 Jazda z prędkością 50 km/h w kierunku stacji Maczki 0.948 M50a A 0.44 0.8 A 0.5 0.65 0.77 0.54 0 5.4 4.8 4.8 Jazda z prędkością 50 km/h w kierunku stacji Maczki A 0.449 0.499 0.0 4.9 5. 8.6 M50a Jazda z prędkością 50 km/h w kierunku stacji Maczki Brak M50a A Figure 7. Accelerometer A. Run at 50km/h toward Maczki Jazda z prędkością 70 km/h w kierunku stacji Jęzor J70a 4.7.76 3.087 J70a A 0.97 0.93.840 0 5.4 4.6 5. 4.6 Jazda z prędkością 70 km/h w kierunku stacji Jęzor A.785.30 A 3.8 4. 4.4 5.3 Jazda z prędkością 70 km/h w kierunku stacji Jęzor.97 0.979 Jazda z prędkością 70 km/h w kierunku stacji Jęzor J70a J70a A.57.40 Brak 5.3 Figure 8. Accelerometer A. Run at 70km/h toward Jęzor 57

EVACES 09 Vibration analysis of bridges Accelerometers (A) were used to measure acceleration in the places of failure which was possible thanks to the way they were placed around the critical places (Figure 4). Figure 7 and 8 show registered with accelometers A and A changes in vertical accelerations a z (t) in time t in place of failure. It pertains to load movements at 50 and 70km/h speeds towards either Maczki or Jęzor. Just as before each picture consist of 4 parts, a to d. Exemplary results are registered with sensor A are show on Figure 7. The vibrograms shown on parts of the pictures show how the vertical acceleration a z changes in time t. With speed 70km/h toward Jęzor, range of vertical accelerations a z () reaches almost, m/s. The measuring signal presented in part was not processed. Registered signal is nonstationary with interferences. It is difficult to conduct vibration analysis for a signal of that kind. Local heaps however can be observed. It is most clearly seen with speeds of 50km/h and 70km/h which reveal F max in ranges 4,9 Hz to 5,5 Hz and 4,8 Hz to, Hz Postprocessing the signal with a filter further increases analysis possibilities. Part of the signal which after post processing of the signal from P sensor with Bessel band filter of 5th order, with lower frequency F d = Hz and top one F g = Hz (graphs b,. Part displays local heap F max in range 0 Hz to 8.6 Hz. With speed 70 km/h a heap appears at 4.9 Hz. In the part F max appears in range.7 Hz to 8.5 Hz while in it takes appears between.7 and 9.5 Hz (in case of the Bessel filter lower frequency is F d =0 Hz and the higher one F g = Hz. Figure 8 shows exemplary results registered with accelerometer A. All values were gathered in Table. Vertical acceleration range a z () were show in column a. Column b, c and d show identified, after processing with filter, frequencies F max corresponding with heaps shown on parts b, c and d of the pictures. 58 Table. Results registered with accelerometers A and A Sensor a z () [m/s ] Identified F max [ Hz] A,377,97 4,8, 0,0 8,6,7 8,5,7 9,5 A,39 4,7 3,3 8,6 3,0 6,4 3,6 8,4 4,6 8,4 Prevailing number of local heaps occurs in vicinity of theoretically established frequencies of natural frequencies for transverse beams next to dilatation (5.7 Hz and 8.5 Hz). Identified frequencies of vibrations F max in range 4.9 Hz to 5.5 Hz most probably connected with further vertical modes of natural vibrations of the main girder. It proves accuracy of a theoretical model for the retrofitted girder. 4.3. Electro resistant tensometers (T) Node number 5 on the left (looking toward Maczki) main girder was chosen for the tests (Figure 3). Figure 4 shows locations of the tensometers (T). 8 of them were used to register changes of the structure deflections, which was made possible by choice of the points the tensometers were placed. During the tests tensometers T and T5 T8 broke down. Exemplary changes of deflections in time registered with tensometers T and T3 are shown on Figure 9 and 0. They pertain to speeds to load movements at 50 and 70km/h speeds towards Jęzor. Just as before each picture consist of 4 parts, a to d. The signal of T, show on Figure 9 is severely interfered. It was impossible to analyze it further. In part one can determine characteristics and range of changes and quasi static deflections during the load runs. In part a heap can be observed at F max =4,58 Hz frequency, however only in part after analysis of 5 s part of registered signal in time, when the load is present on the bridge and the lower frequency of the filter is raised to F d =0 Hz local heaps were revealed for ranges 6,4 Hz to 9,6 Hz. Similar results are obtained for tensometers T3. Also in this case the signal is interfered, but it is possible to determine characteristics of the deflections during the load run. It is further confirmed with post processing with the filter (see part. On both parts (a and clearly can be seen the changes of the tension in bracing. Initial elongation (tension) changes to shortening (compression) with the load run. In part local heaps can be observed with frequencies F max between 4,73 Hz and 8,88 Hz, and in part from 6,0 Hz do 9,4 Hz.

Dynamic tests of certain lattice girder railway bridge type nodes T 794.9 88. 63. 0.03 60.6 T.9 34.7 T 55. 6.0 6. 0 4.58 6.4 6.8 7.8 9.6 95.3 0.0 T Lowpass 5 45.7 49.6 Brak Figure 9. Electro resistant tensometer T. Run at 50km/h. Toward Jęzor T3 7.5 63.8 0.03 0.03 T3 4.3 4.6 8.9 T3 4. 4.4 8.5 0 4.73 7.76 8.88 6.0 8.0 9.4 90.7 Lowpass 5 T3 Brak 5.4 39.3 35.3 Figure 0. Electro resistant tensometer T3. Run at 50km/h. Toward Jęzor 59

EVACES 09 Vibration analysis of bridges All registered values were collected in Table 3. Deflections spans on parts a to d are shown in columns. In further ones identified frequencies F max are shown, after filtering of the signal. Table 3. Results registered with electro resistant tensometers T, T3 and T4 Sensor () [ Str] Identified F max [ Hz] T 63, 95,3 60,6 6, 4,58 6,4 9,6 T3 35,3 90,7 8,9 8,5 4,73 8,88 6,0 9,4 T4 4,5 60, 4,5 8,5 4,58 5, 8,6 Way the tensometers were placed enabled determination of local modes of natural frequencies in places of failure. After filtering according to ( Hz) heaps were identified between 4,58 8,88 Hz (Table 3). They most probably correspond with vertical modes of natural frequencies of the main girder. After filtering (0 Hz) heaps were revealed between 5, 9,6 Hz (Table). These ranges contain theoretical frequencies of couterlaying frequencies and modes of natural frequencies for transverse beams next to dilatations (5,7 Hz and 8,5 Hz). That proves accuracy of the calculation model. 5. SUMMARY It can be observed that found with induction deflection sensors (P) first frequencies agree with theoretical frequencies of nd mode of natural frequencies in both tested spans. Further identification of local modes of natural frequencies was possible with accelometers. In both tested spans prevailing part of heaps was identified close to frequencies corresponding with theoretical natural ones for the transverse beams next to dilatation. That was further confirmed with tensometers which showed heaps close to counter laying theoretical natural frequencies. 6. CONCLUSIONS Induction sensors of deflections (P) allowed registration of main natural vibration modes frequencies for both tested spans. Furthermore accelometers (A) and tensometers (T) allowed winding of local modes of vibrations of the node in both spans. Registered frequencies correspond with theoretical natural vibration modes frequencies. That proves quality of the corresponding theoretical models. Field test provide good basis for analysis of quality of theoretical calculation models. The enable verification of the obtained analysis results. In this case the way the sensors were placed over the structure made possible registration of the natural mode frequencies in the subject spans. Identified first frequencies were identical with theoretical ones. Local frequencies in the tested nodes were identified with tensometers and accelometers. Prevailing part of the heaps was identified next to the theoretical ones which proves quality of the model. In the end that lead to explanation why the cracks appeared in the one of the girders while the second, twin one remained intact. REFERENCES Pradelok, S. (004): The Influence of Higher Modes Vibrations on Local Cracks in Node of Lattice Girders Bridges. PhD Thesis, June 004, Gliwice, Faculty of Civil Engineering, The Silesian University of Technology. 530