Damage Evaluation of Railway Structures based on Train-induced Secondary AE Parameters

Transcription

1 Damage Evaluation of Railway Structures based on Train-induced Secondary Parameters 1 X. LUO, 1 H. HAYA, T. SHIOTANI Railway Technical Research Institute, Tokyo, Japan 1 ; University of Kyoto, Kyoto, Japan Abstract A difficult task for the maintenance of existing railway structures is to nondestructively evaluate the structural integrity of invisible structures as foundations in ground. If visual methods are used, it is generally needed to excavate the subgrade soil surrounding the foundation. As to railway structures this kind of excavation is impractical, because it costs high and also influences the running safety of trains. In the case of such active inspection methods as the ultrasonic testing or X-ray radiography are applied to the foundations, the input power should be strong enough against the decay due to high damping of massive concrete and soil around the foundations, which is also impracticable. Therefore, the authors have developed a passive inspection method taking advantage of secondary (Acoustic Emission) generated at the damage portions of structures, which is induced by trains in service. The train-induced technique, actually, has a potential to evaluate thus damages not only for super-structures but sub-structures since the vast train loading can overcome the high damping caused by the massive concrete of foundation and the surrounding soil. To verify the adequacy of the proposed method, a fundamental study on the characteristics of secondary has been conducted based on experiments of model piles. To improve the applicability of the secondary technique, a lot of in-situ monitoring that took railway bridges as objects have been carried out. Based on the experiments, some practical indices to qualify damage degree of structure, such as RTRI, Caml and Ib-value have been proposed. Introduction Since the 19s many important infrastructures, such as Shinkansen train structures, highways and so forth, have been built in Japan. Because they have been utilized in service for more than years, their maintenance is now becoming significant. Currently, there are many integrity diagnosis methods have been proposed/applied for visible concrete up-structures. For invisible substructures such as foundations, however, effective nondestructive inspection methods are very rare. The main reason to cause this situation is that the most of proposed inspection methods are active ones (e.g. ultrasonic testing, X-ray radiography and impact echo etc.) which need very strong input power to overcome the big decay due to the high damping of massive foundation concrete and surrounding soil. Actually, it is impractical to generate the strong input waves for the active inspections under in-situ condition. Therefore, this is the reason why the authors proposed the train-induced secondary technique [1],[],[3]. inspection is passive as elastic waves are generated due to fracturing or emitted directly from damaged portions. Different from the primary which is based on the Kaiser effect and usually caused by material breaking, the secondary events are generated due to fretting at existing crack surfaces. According to previous studies [4], a seismic diagnosis of concrete piles has been developed by recording the secondary signals induced by large-scale cyclic loading with a vibrator. Moreover, the characteristics of concrete piles suffering from micro-cracking were investigated, and the improved performance of monitoring was confirmed by comparing it with the Pile Integrity Test []. In this paper, to overcome the shortcoming of the method proposed in the reference [4], the large-scale cyclic loading is replaced with train-induced loading and the applicability is examined based on model and in-situ experiments, some concrete piers or columns of existing railway structures with different damage degrees being selected for test purposes. Laboratory experiment with model piles The concept of train-induced secondary inspection is shown in Fig. 1(a). As a newly proposed method, characteristics of secondary were investigated through model pile experiments. Given in Fig. 1(b) is a schematic illustration of the setup and procedure for the experiment with an inverted model pile. To reproduce actual pile behavior, similarity was taken into account while making the specimen. The purpose of the experiment was to investigate the possibility to diagnose pile damage by using the -D (two-dimensional) array composed with the sensors CH1-4 (Fig. 1(b)) set on

2 Step : S im ulated vertical train loading Step 1: S im ulated horizontal loading to damage pile In-service train loading Ground surface Footing Cracks in pier & piles signal : sensor Fig. 1 (a) Proposed inspection; M odel pile φ = 3 Unit: mm 13 Footing Defects due to horizontal loading A E sensors CH1-4 Legend sensor for 3-D monitoring A E sensor for -D monitoring (CH1-4, on surface of footing) Step 3: (i) Detecting defects with array of sensors on surface of footing; (ii) Comparing results from -D and 3-D monitoring. (b) Procedure for experiment with a model pile Pile Footing D Locations -D Locations Elapsed time (sec) (a) Plan view of source locations (b) event rates Fig. Comparison of measurement results between 3-D and -D monitoring the footing surface under the simulated vertical train loading []. To examine the accuracy of the -D monitoring, it was compared with the 3-D results, as illustrated schematically in Fig. 1(b) step 3. For both -D and 3-D measurement, a total of 14 sensors, each with a resonance frequency of khz, were used. Given in Fig. (a) is a plan view comparison of source locations between 3-D and -D monitoring under horizontal loading. The figure shows that the numbers of sources obtained by 3- D (CH1-CH1) monitoring is more than that obtained by -D (CH1-CH4) monitoring. Since the sensor array for 3-D monitoring surrounded the total specimen (pile and footing), all events were well detected. Most of the sources obtained by -D monitoring are located inside the area of pile, which is the expected fractured zone in the vicinity of pile head. Fig. (b) is a comparison between secondary event rates during 3-D monitoring and those of -D monitoring under vertical loading, where events correspond to the number of located sources. Although, the number of events in -D monitoring is less than that of 3-D monitoring, the tendency of generation is quite similar. From the viewpoint of engineering practice, this error level of event for -D monitoring is allowable and the secondary technique is adequate. In-situ monitoring during train operation To verify the implementation of the proposed method, various in-situ monitoring under train loading was carried out, where different kinds of train loading (light, conventional and Shinkansen trains) and structures (piers/columns of viaducts or bridges) are targeted. (1) Light trains and bridge piers The bridge piers were made of plain concrete and constructed more than 7 years ago. Besides the applicability verification, another purpose of the experiment was to investigate the consistency between the proposed method and Impact Vibration Test (IVT) [3]. IVT is an inspection method to evaluate the damage degree of a pier by measuring its natural frequency that is caused by impacting event rate (Hit / sec) 1 3-D monitoring -D monitoring

3 Fig. 3 Whole scheme and photos of objective bridge with pier damage evaluation results of IVT the pier with a heavy spindle. This method has been widely used for maintenance of railway structures in Japan. Fig. 3 shows a whole scheme and some photos of objective bridge with pier damage evaluation results of IVT. According to maintenance record, this bridge was intensely shaken by an earthquake occurred previously. The inspection results indicated in Fig. 3 show that the P3 and P4 piers had suffered high degree damage, in contrast the P only suffered low degree damage. Besides the monitoring, in the experiment some other measurements for strain, crack, Strain gauge Strain gauge 1 Acceleration sensor acc S S Legend sensor π-shape gauge 1 π π π-shape gauge for displacement Construction joint Visible damage zone π-shape gauge π π S acc Strain gauge Acceleration sensor π-shape gauge 3 Fig. 4 Schematic illustration of various sensors setting on P3 pier

4 Bridge pier Bridge pier Bridge pier Vertical (m) 4. Ch-4 Ch-3 Ch- Event number: 4 Ch-8,1 Ch-7,11 Ch-, Vertical (m) 3. Event number: 84 Vertical (m). 4. Event number: Cross track direction (m) - Ch-1 Construction join Ch-,9 - Elevation Ch-1~4 Crack Ch-~8 Construction join - Elevation P3 P4 P Cross track direction (m) Crack Cross track direction (m) - Elevation Ch-9~ Plane Plane Plane Legend : sensor; : Location of source (the size of circle represents the counter of wave) Fig. Secondary sources located by 3-D monitoring corresponding to each pier acceleration and train loading were simultaneously conducted to grasp the relationships among generation, pier deformation and loading action. A schematic illustration of various sensors setting on the P3 pier is taken as an example shown in Fig. 4. In the monitoring, 1 sensors with resonant frequency of khz were used. Fig. shows the monitoring results induced by the light train loading corresponding to the three piers. Under same measurement condition, the event numbers and the counters of waves are obviously different between the piers with high damage degree (P3 and P4) and low damage degree (P). In contrast with the event numbers of 4 and 84 corresponding to the P3 and P4 piers, there are only events for the P pier. Moreover, the counters of waves obtained in the P3 and P4 piers are larger than those in the P in average. As a result, the characteristics of event measured are coincident with the damage evaluation by the IVT. () Conventional trains and arch-shaped piers Another in-situ monitoring was performed focusing on the relationship between the damage degree and the loading/unloading activities. The object for the experiment was an arch-type pier, which was made of plain concrete and built more than 7 years ago. A schematic illustration and Train loading signal sensor. Down train side 3.3 Unit:m 3. Cracks due to aging.7 Up train side Fig. (a) Schematic illustration of monitoring; 3.3 (b) Photos of pier with severe crack zone

5 photo showing the monitoring of the pier utilizing the loads of operating trains, are shown in Fig.. The surface cracks in the pier, the arrangement of the sensors, and locations of the secondary events determined are shown in Fig. 7. In the figure, 1 ( khz resonant) sensors were arranged three-dimensionally around the damaged zone, source results showing that most of the signals emanated from the fractured zone. According to previous studies [7], indices called the Load ratio (Ratio of load at the onset of activity to previous load) and Calm ratio (Ratio of cumulative activity under unloading to that of previous maximum loading cycle) were applied to evaluate the structural damage, which is based on the concept of the Kaiser effect. In the case of secondary, however, this concept is not readily applicable and. furthermore, since it is difficult to estimate the maximum load that has ever been experienced by the pier during earthquakes, and the live loads due to trains in service, the calculation of Load ratio is impractical. Therefore, a new index called the RTRI ratio (ratio of Repeated Train load at the onset of activity to Relative maximum load for Inspection period) is being proposed. This modifies the definition of Load ratio by introducing the relative maximum load instead of the previous maximum one. If the load due to train passage is difficult to measure in situ, the concept of load can be replaced by deformation of structure, which reflects load variations and is easy to be measured in situ. In the experiment, the RTRI ratio was determined by using the train-induced displacement at Crack 3 shown in Fig. 8. The displacement was confirmed to have a close relation with the train loading by comparing with riding density of the train passed. Therefore, the histories of displacement were adopted to represent the variation of the train-induced live loads to the pier. The formulae for calculating the RTRI and Calm ratios are given below and the calculation example is shown in Fig. 8. Displaceme ntonset of activity RTRI = Displaceme ntmax load during inspection period (1) Cumulative Activity from peak of Calm = Cumulative Activity from beginning of displaceme nt to end displaceme nt to peak () A Cracks or repair traces in back Sensor in back 9 11 Cracks or repair traces in front Sensor in front Sensor in back A Elevation A-A 9 11 Legend : source : sensor No.: Number of sensor Plan Fig. 7 Secondary sources located by 3-D monitoring

6 Cumulative hits 1 Displacement increase Displacement at onset of activity, 71mm 143 Hit 791 Hit Displacement decrease 1:9:3 17:: 17::3 Elapsed time (sec) Relative max displacement.11 mm 171 Hit Cumulative hit Unloading 398 Hit RTRI = = Displacement of crack (mm) Loading Hit Calm = =.73 4 History of displacement Fig. 8 Calculation of the RTRI and Calm ratios based on variation of displacement at Crack 3 Calm ratio Damage degree high Damage degree middle Damage degree middle Damage degree low RTRI ratio Fig. 9 Damage qualification of pier based on RTRI ratio and Calm ratio The qualified damage degree of the pier based on the RTRI and Calm ratios obtained above is demonstrated in Fig. 9. The results show that most of the measured data were plotted within the Damage degree high zone (Calm ratio >, RTRI ratio <.8). This is reasonably in agreement with the actually damaged situation of the pier. Therefore, the proposed RTRI ratio index, based on secondary activity is proved to be both useful and suitable for in-situ monitoring. The limits to qualify the damage degree in Fig. 9 were determined after evaluating the real damage of the pier with other NDI methods [3]. (3) Shinkansen trains and viaduct columns Given in Fig. is the illustration for one of the in-situ experiments where a column of viaduct for Shinkansen train was damaged by an earthquake in 3. From the situation of crack density and width which were investigated after the earthquake, the damage zones can be divided as high degree zone (close to the footing) and middle degree zone (above the high degree zone). To compare the characteristics due to the zones with different damage degrees, the sensors were set up as Array A in the zone with high damage degree, Array B in the zone with middle damage degree (Fig.(a)). Fig.(b) shows the secondary sources located by the each array under in-service Shinkansen train loading. Since the circle size of source shown in Fig.(b) represents the counter of signal, it is revealed that the number and intensity of event corresponding to the Array A with severer damage are much bigger than those corresponding to Array B with middle level damage.

7 Column Crack width mm Damage degree middle Array B a 4 Array B Array A a Crack width 1.4mm Damage degree high Array A Footing South (#11) #9 (#1) # a-a # (#4) #1 (# 3) # (#8) North # (#7) (a) Damage column with sensor Array A (damage degree high) & B (damage degree middle) X (m) Z (m) Array A Z (m) (b) 東側面 South North 9 1 Y (m) (d) 平面 Plan 4 Up trains Down trains West (b) Secondary sources located by Array A & B Fig. Damage state of objective column and monitoring results corresponding to Array A & B X (m) Z (m) Array B Z (m) (b) 東側面 South 9 1 North Y (m) Plan (d) 平面 4 West

8 In addition to the indices of RTRI and Calm shown above, peak amplitudes are also known to be closely related to the scale of fracture, namely damage degree. Thus, the scale of peak amplitude would be larger with progress of fracture. Many researchers have studied b-value determined from a negative gradient of the peak-amplitude distribution. To apply the b-value analysis for the evaluation of stability in slopes, the method for calculating the b-value has been modified by incorporating statistical values of amplitude distribution []. This is now referred to as improved b-value (Ib-value) analysis. In this experiment, the relationships between structural damage degree and the Ib-value were also investigated based on the in-situ monitoring. The results of Ib-value corresponding to Array A and B are shown in Fig. 11. In the figure, the open circles (Array B) and open squares (Array A) show the differential type of distribution, the solid symbols show the cumulative type of distribution, and the cross symbols represent the average values of peak amplitudes corresponding to the both arrays. It is found that the solid plots show that the gradient in Array A ( ) appears smaller than that in Array B (.). This implies that the Ibvalue becomes smaller as the damage degree increases. Moreover, by considering Ib-values obtained from other in-situ monitoring, it might be suggested to determine the Ib-value of around as a limit for qualifying severe damage. Conclusions Number of events (Differnetial frequency) Average in B Array array B Average in in A Array array A : A Array arraya : Array B arrayb Array A-array A y = -9x R =.9739 Array B-array B y = -94x R = Peak amplitude (db) Fig. 11 Amplitude distribution for Array A and B To develop an efficient and applicable method for integrity diagnosis of invisible structures, the secondary technique is investigated in this study. As a newly proposed method, firstly, the basic characteristics of secondary are investigated through the model pile experiments. The experiments testify the possibility to diagnose pile damage by using the -D (two-dimensional) array set on the footing surface under the simulated vertical train loading Secondarily, the in-situ monitoring is conducted with different types of structures (RC piers and columns) and train passages (light, conventional and Shinkansen trains) and the following conclusions are obtained. 1) The secondary events generated by fretting at surfaces of existing cracks are detectable under train loading. The event numbers and the counters of waves are obviously different between the structural zones with high damage degree and low damage degree. ) By comparing the results obtained with the proposed technique and the Impact Vibration Test (IVT), the consistency in damage qualification between the two methods is proved. 3) Structural damage degree can be quantitively evaluated with the indices RTRI, Calm and Ib-value that are based on the parameters and structural deformation measured in situ. Accumulated number of events (Cumulative frequency)

9 Since the proposed method takes advantage of large loading due to in-service trains, the cost for inspection is much lower than the active NDI methods. Also the method is suitable to detect invisible structures, which is a most difficult problem in maintenance of railway structure. Moreover, its implementation has been verified through various in-situ experiments and the good cost & performance to railway business is obvious. With regard to future studies, the reliability of the proposed method will be improved. References [1] X. Luo, H. Haya, T. Inaba, T. Shiotani and Y. Nakanishi. Study on the applicability of Traininduced secondary for nondestructive inspection of structures, Proc. 1 th International Symposium, Tokushima, Japan. pp.-1, (). [] X. Luo, H. Haya, T. Inaba, T. Shiotani and Y. Nakanishi. Damage evaluation of railway structures by using train-induced, International Journal Construction and Building Materials, Vol.18, No.3, pp.1-3, (4). [3] X. Luo, H. Haya, T. Inaba and T. Shiotani. Seismic diagnosis of railway substructures by using secondary acoustic emission ; International Journal Soil Dynamics and Earthquake Engineering, Vol., No.1, pp.11-11, (). [4] T. Shiotani, N. Sakaino, M. Shigeishi, M. Ohtsu, Y. Asai and T. Hayashi. Characteristics of full-scale concrete-pile under bending and shear load, Proc. th International Symposium on Acoustic Emission from Composite Materials, pp (1998). [] T. Shiotani, M. Shigeishi and M. Ohtsu. Acoustic emission characteristics of concrete-piles, International Journal Construction and Buildings Materials, 13:pp.73-8, (1999). [] T. Shiotani and M. Ohtsu. Prediction of slope failure based on activity, Acoustic Emission: Standards and Technology Update, ASTM STP, 133:pp.1-17, (1999). [7] Recommended practice for in situ monitoring of concrete structures by acoustic emission, Japanese Society for Nondestructive Inspection (NDIS), NDIS 41, (in Japanese), ().