Publications

Detailed Information

Enhancing damping estimation in cable-supported bridges under operational conditions : 공용중 케이블교량의 계측데이터 기반 감쇠비 추정기법 개선

DC Field Value Language
dc.contributor.advisor김호경-
dc.contributor.author김선중-
dc.date.accessioned2017-10-27T16:31:27Z-
dc.date.available2017-10-27T16:31:27Z-
dc.date.issued2017-08-
dc.identifier.other000000145569-
dc.identifier.urihttps://hdl.handle.net/10371/136690-
dc.description학위논문 (박사)-- 서울대학교 대학원 공과대학 건설환경공학부, 2017. 8. 김호경.-
dc.description.abstract응답기반모드해석(Operational Modal Analysis, OMA) 기반 추정 감쇠비의 불확실성을 줄이기 위해 자동화된 변수선정 및 신호 정상화 과정에 기반한 개선된 Natural Excitation Technique—Eigensystem Realization (NExT-ERA) 알고리즘을 제안하였으며, 다중질량동조감쇠기(Multiple Tuned Mass Damper, MTMD) 설치 전후의 사장교 계측 데이터에 적용하여 그 정당성을 검증하였다. 상관분석을 통해 환경적 요인의 영향을 분석하고 감쇠비 변화에 가장 주요한 원인이 되는 환경적 요인을 선정, 회귀분석을 수행하여 제안한 방법 적용 전후에 따라 감쇠비의 불확실성이 얼마나 줄어드는지를 확인하였다.
경험적 변수선택에 준하는 감쇠비 추정의 정확도 및 정밀도의 확보를 목표로 자동화된 변수 선정 절차를 제안하였다. 먼저 적절한 변수 구간을 선정하기 위해 수치해석과 계측 데이터를 활용한 변수연구를 수행하였다. 평균과 변동계수를 통해 정확도 및 정밀도를 평가한 결과, 샘플링 주파수 100Hz에서 데이터 길이 60분에 Number of FFT 215 일 때 가장 좋은 결과를 보였다. 행켈 행렬의 크기는 계산된 임펄스응답함수가 최대값의 50%만큼 되었을 때를 기준으로 결정하였다. 시스템 차수에 따른 추정 감쇠비의 분산을 해결하기 위해, 시스템 차수를 1부터 50까지 바꿔가며 추정한 감쇠비의 중앙값을 취하였다. 상기의 자동화된 변수 선정 절차는 경험적 변수선택보다 더욱 정확한 결과를 보였다.
현장실험을 통해 차량으로 인해 응답이 집중화되는 것과 그것이 OMA 기반 감쇠비 추정에 미치는 영향을 분석하였다. 교량의 차선이 적고 통행량이 많지 않을 경우, 차량이 센서에 근접함에 따라 상시계측 가속도가 커지고 차량이 센서를 지나감에 따라 응답이 줄어들게 된다. OMA는 하중을 정상상태 백색잡음으로 가정하기 때문에, 이러한 차량에 의한 하중 및 응답의 비정상성은 OMA 불확실성의 한 원인이 된다. 또한 이러한 차량 하중은 계측 가속도의 파워스펙트럼함수(Power Spectral Density function, PSD)의 주파수 성분을 왜곡한다. 정상상태 하중 및 차량하중에 의한 계측가속도 PSD를 비교했을 때, 차량이 지나갈 때 2-5Hz의 고주파 성분이 증폭되는 것을 확인하였다. ERA는 에너지 기반의 기법으로, 지배적인 모드를 보다 쉽게 식별하기 때문에 이러한 PSD의 왜곡 역시 추정 감쇠비에 오류를 야기할 수 있다.
계측 데이터의 시계열에서 나타나는 비정상성을 제거하기 위해, 진폭변조함수를 통한 신호 정상화 과정을 제안하였다. 비정상성 시계열은 포락함수와 정상 시계열의 곱으로 표현할 수 있다. 따라서 계측데이터의 이동 Root-Mean-Square(RMS)로부터 포락함수을 구해 계측 데이터를 포락함수로 나누면 계측 데이터 내 정상 시계열만을 성공적으로 추출할 수 있다. 제안한 신호 정상화 과정을 3일간의 상시진동 데이터에 적용하여 감쇠비를 추정한 결과 추정치의 큰 편향들이 제거되었고, 변동계수 역시 줄어들었다. 이를 통해 신호 정상화 과정이 감쇠비의 불확실성을 제거하는데 유효함을 확인하였다.
대상교량에 MTMD가 설치된 이후의 감쇠비를 동일한 방식으로 추정하였다. 다양한 바람 조건 하에서 계측된 4일간의 데이터를 통해 감쇠비를 추정한 결과, 역시 자동화된 변수 선택 및 신호 정상화 과정이 OMA 기반 추정 감쇠비의 불확실성을 줄여주는 것을 확인하였다. 특별히 MTMD에 의한 감쇠비 증가 효과는 풍속이 와류진동 위험풍속일 경우 가장 컸으며, 일반적인 풍속 조건에서는 비록 설계수준의 성능을 보이지는 않았음에도 불구하고 충분한 제진 효과를 나타내었다.
마지막으로 환경적 요인이 감쇠비 변화에 미치는 영향을 평가하였다. 진폭, 차량 대수 및 온도를 대상 변수로 선정, 상관성 분석을 통해 각 환경적 요인과 감쇠비 사이의 관련성을 분석하였다. 그 결과 차량에 의한 진동이 케이블 교량 진동 발현에 가장 큰 요인임을 확인하였다. 와류진동 Lock-in 구간에서도 진폭이 증가하였으나, 이는 감쇠비 증가에는 영향을 미치지 않는 것으로 나타났다. 온도와의 상관성 역시 상대적으로 낮았다. 대표 변수인 차량에 의한 진폭 의존도(amplitude-dependency)를 회귀분석을 통해 분석한 결과, 신호 정상화 과정을 적용하기 전에는 거의 나타나지 않던 상관성이 신호 정상화 적용에 따라 명확하게 나타나는 것을 확인하였다.
-
dc.description.abstractAn enhanced output-only algorithm of the Natural Excitation Technique—Eigensystem Realization Algorithm (NExT-ERA) was suggested to solve a large scattering in identified damping ratio still remains to be a challenging issue. The suggested damping estimation procedure was applied to a parallel cable-stayed bridge for identifying structural damping ratios before and after the installation of a Multiple Tuned Mass Damper (MTMD) designed to mitigate a vortex-induced vibration (VIV) that was observed on the bridge.
The automated proper parameter selection process for NExT-ERA was suggested to reduce the large error bound due to poor parameter selection and achieve similar level of correctness of heuristic parameter selection. To make suitable criteria for each algorithm parameter, a series of parametric studies using numerical simulation and operational monitoring data was performed, respectively. It was discovered that the number of FFT (NFFT) of 215 with 60 min data provided the accurate estimation in terms of mean and coefficient of variance (COV). A size of Hankel matrix was determined corresponding to the shape of calculated impulse response function (IRF). To overcome the limitation of non-structural model based algorithm, the sensitivity analysis was performed using each estimated value according to a system order. The median value of estimated damping ratios provided a converged value successfully. This automated parameter selection process accomplished the more accurate damping estimation compared to the result of heuristic parameter selection.
The research also discovered that the effect of traffic loadings on the uncertainty of operational modal analysis (OMA) based damping estimation. The experimental studies found a localized response in traffic-induced vibration (TIV). When the number of traffic lane over the bridge is only one or two for one direction and the traffic volume is not high, the ambient vibration signal at the sensor position show an envelope as a vehicle is approaching and fading away. Since OMA assumes that the signal to be analyzed is a stationary white-noise process, the envelop-like signal obtained from running vehicles can also contribute the scattering in Structural Identification (St-Id). Furthermore, this traffic loading distorted the PSD of measured acceleration. A comparison between the PSD of stationary response and TIV clearly demonstrated that the frequency components of 2-5 Hz were amplified during the vehicle crossed the bridge. The ERA is the energy-based method so that the dominant modes will be more easily identified. Therefore, the distortion in PSD can be a reason for the poor modal identification.
To remove this envelope-like trend in measured data, a signal stationarization process based on amplitude-modulating function was employed to the simulated response and measured data, respectively. If the nonstationary loading can be represented by the product of amplitude-modulating function and stationary white noise process, then the envelope function can be evaluated by temporal root-mean-square function of responses. Consequently, the approximated stationary process can be extracted by dividing the measurements with calculated envelope function. This signal stationarization process successfully extracted the stationary process from the TIV. This signal stationarization process was applied to the 3-day operational monitoring data. It is discovered that some highly scattered points were eliminated by the signal stationarization. The COV of estimated damping ratios is significantly reduced, indicating the signal stationarization process is worth in reducing scattering in identified damping ratios from OMA. As a result, the amplitude dependency of a damping ratio also more clearly appeared in terms of R-squared value of linear regression increasing.
Since a multiple tuned mass damper (MTMD) was installed at the center of the main span to mitigate VIV, the modal damping ratios before and after the installation of the MTMD were also compared. Several sets of operational monitoring data that had been collected under various windy conditions were used to develop a relationship between the identified damping ratios and the corresponding VIV level of the bridge. The performance of the bridge was enhanced regarding vibrational serviceability, based on the above findings.
The effect of environmental factors on the variation of damping ratio was evaluated. Three environmental factors of vibration amplitude, the number of vehicle and temperature were selected, and the statistical relationship between the damping ratio and each environmental parameter was evaluated through correlation analysis. The result confirmed that the main source of damping ratio of cable-supported bridge was traffic-induced vibration which showed a high positive correlation with the number of vehicle and corresponding vibration amplitude of TIV. RMS amplitude also increased within lock-in range, but no corresponding increase in the damping ratio was observed. The effect of temperature changes was also relatively low. The analysis of the amplitude-dependency of the damping ratio clearly showed that a tendency was clearly appeared by applying the stationarization method to reducing uncertainties in damping estimates.
-
dc.description.tableofcontents1. Introduction 1
1.1. Research background 3
1.2. Problem definition 7
1.3. Objective and scope 9
2. Damping estimation based on operational modal analysis 13
2.1. NExT-ERA: Output-only operational modal analysis 15
2.1.1. Literature survey: OMA-based damping estimation 15
2.1.2. Natural Excitation Technique (NExT) 17
2.1.3. Eigensystem Realization Algorithm (ERA) 23
2.2. Damping estimation of cable-stayed bridge 29
2.2.1. Bridge description: Jindo Bridge 29
2.2.2. Monitoring data 30
2.2.3. Excitation test using TMD 33
2.2.4. Operational damping estimation using 3-days data 38
2.3. Concluding remarks 42
3. Automated proper parameter selection for NExT-ERA 43
3.1. Analysis parameter of NExT-ERA 45
3.2. Parametric studies of numerical simulation 46
3.2.1. Description of the simulated models 46
3.2.2. Algorithmic parameters of NExT: NFFT and data length 47
3.2.3. Algorithmic parameters of ERA: Size of Hankel matrix and system order 53
3.3. Parametric studies using field monitoring data 58
3.3.1. NFFT and the record length 59
3.3.2. Size of Hankel matrix 60
3.3.3. System order 61
3.4. Application: Jindo Bridge 63
3.4.1. Parameter selection 63
3.4.2. Estimated damping ratio after proper parameter selection 66
3.4.3. Computational cost 68
4. Signal stationarization for traffic-induced vibration 71
4.1. Introduction 73
4.2. Experimental investigation for TIV properties 75
4.2.1. Bridge description: Sorok Bridge 75
4.2.2. Monitored data 76
4.2.3. Experimental condition 77
4.2.4. Nonstationary effect of TIV on a measured signal 80
4.3. Signal stationarization using amplitude-modulating function 85
4.3.1. Amplitude-modulating function 85
4.3.2. The optimal segment length for signal stationarization 87
4.3.3. Signal stationarization of operational monitoring data 91
4.4. Application to NExT-ERA: numerical simulation 93
4.4.1. Description of simulated models 93
4.4.2. Estimated damping ratio corresponding to signal stationarization 100
4.5. Application to NExT-ERA: operational monitoring data of Jindo Bridge 102
4.5.1. Combined framework of p.p.s and stationarization 102
4.5.2. Monitoring data 104
4.5.3. Effect of signal stationarization in Jindo Bridge 104
4.5.4. Estimated damping ratio according to signal stationarization 107
4.6. Application to NExT-ERA: Jindo Bridge after installation of MTMD 112
4.6.1. Monitored data 112
4.6.2. Estimated damping ratio corresponding to signal stationarization 114
4.6.3. Effectiveness of MTMD according to windy condition 118
4.7. Concluding remarks 126
5. Environmental effect on the variation of estimated damping ratio 127
5.1. Main environmental factors for variation of damping ratio 130
5.1.1. Vibration amplitude 130
5.1.2. Number of vehicle 132
5.1.3. Temperature 133
5.1.4. Aerodynamic damping ratio 135
5.2. Correlation analysis between estimated damping ratio and environmental factors 138
5.3. Correlation analysis between estimated damping ratio and environmental factors without lock-in range data set 141
5.4. Amplitude-dependency 143
5.5. Conclusion 146
6. Conclusions and further study 147
6.1. Conclusions and contributions 149
6.2. Suggestion 151
References 154
-
dc.formatapplication/pdf-
dc.format.extent4702017 bytes-
dc.format.mediumapplication/pdf-
dc.language.isoen-
dc.publisher서울대학교 대학원-
dc.subjectOperational Modal Analysis-
dc.subjectDamping ratio-
dc.subjectNExT-ERA-
dc.subjectCable-supported bridge-
dc.subjectStationarization-
dc.subjectAmplitude-dependency-
dc.subject.ddc624-
dc.titleEnhancing damping estimation in cable-supported bridges under operational conditions-
dc.title.alternative공용중 케이블교량의 계측데이터 기반 감쇠비 추정기법 개선-
dc.typeThesis-
dc.description.degreeDoctor-
dc.contributor.affiliation공과대학 건설환경공학부-
dc.date.awarded2017-08-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share