Geri Dön

Bina tipi taşıyıcı sistemlerde bir yapı sağlığı izleme yöntemi önerisi: Basitleştirilmiş sayısal model ve hasar tespit algoritması

A structural health monitoring method proposal for building-type structural systems: Simplified numerical model and damage detection algorithm

PDF İndir

  1. Tez No: 995864
  2. Yazar: EMRE GÖNÜLCÜ
  3. Danışmanlar: PROF. DR. ENGİN ORAKDÖĞEN, DR. ÖĞR. ÜYESİ AHMET ANIL DİNDAR
  4. Tez Türü: Doktora
  5. Konular: İnşaat Mühendisliği, Civil Engineering
  6. Anahtar Kelimeler: Yapı dinamiği, Structural dynamics
  7. Yıl: 2026
  8. Dil: Türkçe
  9. Üniversite: İstanbul Teknik Üniversitesi
  10. Enstitü: Lisansüstü Eğitim Enstitüsü
  11. Ana Bilim Dalı: İnşaat Mühendisliği Ana Bilim Dalı
  12. Bilim Dalı: Yapı Mühendisliği Bilim Dalı
  13. Sayfa Sayısı: Belirtilmemiş.

Özet

Bu tez çalışmasında, çok katlı yapıların deprem sonrası yapısal davranışlarının değerlendirilmesi ve olası hasar bölgelerinin belirlenmesi ve derecelendirilmesi amacıyla titreşim tabanlı bir Yapı Sağlığı İzleme yöntemi geliştirilmiştir. Türkiye gibi deprem riski yüksek bölgelerde, yapı güvenliğinin hızlı ve güvenilir biçimde değerlendirilebilmesi büyük önem taşımaktadır. Bu kapsamda çalışma, sayısal modeller kullanılarak olası tüm hasar durumlarına ait senaryoların oluşturulması ve deprem sonrasında ölçülen dinamik tepkilerle bu senaryolardan hangisinin gerçekleştiğinin belirlenmesi esasına dayanmaktadır. Yöntem, geliştirilen Basitleştirilmiş Sonlu Elemanlar Modeli (BSEM) ve Hasar Duyarlılık Matrisi (HDM) yaklaşımlarını bir araya getirerek süreci önemli ölçüde hızlandıran bir akış şeması önermektedir. Tezin ilk bölümünde, literatürde yer alan klasik YSİ teknikleri değerlendirilmiş; zaman, frekans ve mod tabanlı sistem tanımlama yaklaşımları incelenmiş ve özellikle geleneksel ve yenilikçi yöntemlerin avantajları ve kısıtları değerlendirilmiştir. Bu kapsamda, İstanbul'da yer alan 53 katlı bir yüksek yapı örnek olarak ele alınmış ve geleneksel YSİ yaklaşımları bu yapı üzerinde uygulanarak sistem tanımlama ve model kalibrasyonu gerçekleştirilmiştir. Böylece yüksek binalarda modal tanımlama ve model kalibrasyonu süreci gerçek bir yapı üzerinde detaylıca ortaya konmuştur. Çalışmanın ikinci kısmında, klasik FEM modellerinin karmaşıklığı ve yüksek hesaplama maliyetleri dikkate alınarak Basitleştirilmiş Sonlu Elemanlar Modeli (BSEM) olarak adlandırılan yeni bir modelleme yaklaşımı geliştirilmiştir. Bu model, yapının genel dinamik davranışını koruyarak eleman sayısını azaltmakta ve analiz süresini önemli ölçüde düşürmektedir. BSEM kapsamında iki farklı kalibrasyon tekniği önerilmiştir: bunlardan ilki yalnızca birinci moda dayalı tek modlu kalibrasyon, diğeri ise birden fazla moda ait verilerin eşzamanlı olarak kullanıldığı çok modlu kalibrasyon yöntemidir. Bu iki yaklaşım 7 katlı gerçek bir bina üzerinde sayısal olarak test edilmiştir. Analizler sonucunda, basitleştirilmiş modelin karmaşık sonlu eleman modellerine yakın doğrulukta sonuçlar verdiği, bununla birliktr işlem süresini önemli ölçüde kısalttığı belirlenmiştir. Tezin üçüncü kısmında, yapısal hasarın yerini ve büyüklüğünü belirlemek amacıyla geliştirilen Hasar Duyarlılık Matrisi (HDM) kavramı ayrıntılı olarak ele alınmıştır. HDM, Bina tipi sistemlerde kat bazında rijitlik kaybı konumunun modal periyotlar üzerindeki etkisini ifade eden bir duyarlılık göstergesidir. Yöntemin oluşturulmasında öncelikle yapının hasarsız durumu için kalibre edilmiş sayısal bir model hazırlanır, ardından her bir kat için ayrı ayrı olacak şekilde farklı hasar seviyelerinde rijitlik kayıpları tanımlanarak olası tek kattaki tüm hasar durumları modellenir. Her hasar seviyesi için yapılan modal analizler sonucunda, modal periyotlarda meydana gelen değişimler belirlenir ve bu değişim oranları matris biçiminde düzenlenerek HDM oluşturulur. Böylece her hücre, belirli bir kattaki belirli bir hasar düzeyinin modal parametreler üzerindeki etkisini temsil etmektedir. Deprem sonrası yapıya ait ölçülen titreşim verileri kullanılarak elde edilen yeni modal periyotlar, HDM'den üretilen sayısal senaryolar ile karşılaştırılır ve en uygun eşleşme üzerinden hasarın yeri ile büyüklüğü belirlenir. Bu yöntem karmaşık optimizasyon süreçlerine veya yüksek hesaplama maliyetine sahip doğrusal olmayan analizlere gerek kalmadan, kısa sürede güvenilir hasar tespiti yapılmasına olanak sağlar. Nümerik, deneysel ve gerçek yapı üzerinde yapılan çalışmalar, önerilen yöntemin hem doğruluk hem de hız bakımından etkili performans gösterdiğini ortaya koymuştur. Ayrıca mod şekillerinin normalize edilmiş halleri kullanırlarak HDM'lerin analitik olarak tahmin edilmesine olanak tanıyan ilişkinin ilk adımları paylaşılmıştır. Sonuç olarak, bu tezde geliştirilen yöntem bina tipi yapılarda deprem sonrası yapısal hasarların belirlenmesi için yenilikçi bir yaklaşım sunmaktadır. Basitleştirilmiş Sonlu Elemanlar Modeli (BSEM), Hasar Duyarlılık Matrisi (HDM) ve senaryo tabanlı analiz yapısı sayesinde süreç hem hızlandırılmış hem de güvenilir hale getirilmiştir. Önerilen yöntem, gerçek binalara uygulanabilir, ölçeklenebilir ve sensör tabanlı izleme sistemleriyle entegre niteliktedir.

Özet (Çeviri)

In this doctoral research, a comprehensive and innovative vibration-based Structural Health Monitoring (SHM) methodology has been developed for the post-earthquake evaluation of multi-story building-type structures. The primary goal of the study is to establish a reliable, rapid, and computationally efficient framework for identifying, localizing, and quantifying structural damage using ambient or earthquake-induced vibration data. Rapid assesment after earthquakes is crucial for minimizing human and economic losses, especially in seismically active regions such as Turkey, the proposed methodology aims to fill the gap between complex analytical models and practical monitoring systems applicable to real structures. The study integrates a new modeling technique called Simplified Finite Element Modeling (BSEM) and a new methodology for damage localization and quantification in multi-story buildings called Damage Sensitivity Matrix (HDM). The research begins with an extensive review of the literature on vibration-based SHM methods. Over the past two decades, numerous studies have explored vibration-based identification techniques, including time-domain, frequency-domain, and hybrid approaches, to detect changes in structural stiffness that may indicate damage. However, many of these methods either rely heavily on dense sensor networks or demand extensive computational resources for model updating, making them impractical for large-scale or real-time applications. Moreover, conventional finite element models, while highly detailed, often involve thousands of degrees of freedom and require significant computational effort, preventing their use for rapid postearthquake assessments. In contrast, purely data-driven methods such as those based on artificial intelligence or statistical pattern recognition. This research proposes a new, extremely fast-producible finite element method that combines the physical dimensions of the structural system with measurement data obtained from the SHM system. Furthermore, this modeling technique, along with a damage sensitivity matrix method that can rapidly generate all possible damage scenarios, creates a framework that provides rapid and reliable information on the location and extent of postearthquake damage. Before introducing the proposed modeling technique and damage detection algorithm, system identification techniques such as modal frequency, damping ratio, and mode shapes are presented within the context of vibration-based SHM. To investigate these techniques, a 53-story twisted-form skyscraper in Istanbul was instrumented with accelerometers. Data collected continuously over the past three years (ambient vibration and small-scale earthquake records) were used to determine the dynamic characteristics of the structure. In addition to these data, the behavior of the sample building during the Mw 6.2 Istanbul earthquake on April 23, 2025, was investigated in detail. Finally, in this part of the study, the numerical model developed for the sample building was calibrated, and the effects of the effective section stiffness and dynamic modulus of elasticity on model calibration were examined. The conventional finite element modeling and analysis of complex structures are often computationally intensive and time-consuming. To overcome this limitation, a Simplified Finite Element Model (BSEM) was developed as a core component of this study to minimize computational demand during repetitive analyses and to rapidly predict the post-earthquake response of floors without sensors. The BSEM approach considerably reduces the number of elements and overall computational cost while maintaining the essential stiffness, mass, and damping characteristics of the original system. In this formulation, the building is idealized as a series of lumped masses interconnected by equivalent stiffness elements along its height, enabling the simulation of global dynamic behavior with a minimal parameter set. The performance of the BSEM was validated through numerical analyses of four tall buildings with varying geometric configurations and stiffness distributions. Comparisons between detailed and simplified finite element models confirmed the accuracy and efficiency of the proposed methodology. A key feature that distinguishes this method from those reported in the literature is its ability to incorporate torsional modes into the analysis, thereby providing a more comprehensive representation of the building's dynamic behavior. Two calibration strategies were proposed: (1) a single-mode calibration, which utilizes the dominant vibration mode, and (2) a multi-mode calibration, which incorporates multiple modes to enhance precision. Both strategies were tested using experimental data from a seven-story reinforced concrete Van Nuys building. The findings demonstrated that the BSEM can reproduce the modal frequencies and mode shapes of detailed models with high fidelity while requiring significantly less computational time, making it suitable for real-time or near-real-time SHM applications. Although the multi-mode calibration provides higher accuracy, it is moderately more time-consuming; therefore, the choice between methods can be made based on cost–efficiency considerations. The most significant contribution of this study is development and implementation of the Damage Sensitivity Matrix (HDM), which constitutes the analytical core of the proposed damage detection framework. The HDM quantifies the relationship between localized stiffness reductions on each floor of the building and the corresponding variations in modal periods. To construct the HDM, a calibrated undamaged BSEM was initially developed. Predefined stiffness reduction ratios (typically 20%, 40%, 60%, and 80%) were sequentially introduced to each floor, and modal analyses were carried out for each simulated damage scenario. The resulting variations in natural periods were systematically documented, forming a matrix in which each element denotes the sensitivity of a given vibration mode to stiffness loss at a specific floor. Unlike conventional sensitivity-based approaches, the HDM enables a rapid and systematic assessment of damage distribution along the height of tall buildings, providing an efficient foundation for post-earthquake damage evaluation. During post-earthquake evaluation, the newly measured modal parameters from vibration data are compared with the pre-established HDM database. The most probable damage scenario is then identified by applying a minimum-error matching criterion, which quantifies the difference between the measured and simulated modalperiods. This approach eliminates the need for complex optimization algorithms and provides a rapid, transparent, and physically interpretable means of localizing and quantifying damage. Furthermore, since the HDM is derived from a simplified yet accurate model, the analysis can be performed almost instantaneously after new data become available, offering a practical solution for emergency response and safety evaluation in the aftermath of an earthquake. The proposed formulation enables the incorporation of individual story-level damage effects into the damage matrix, allowing for the consideration of all possible combinations of damage scenarios. During post-earthquake evaluations, the newly measured modal periods derived from vibration data are compared with the estimated periods obtained from the preestablished HDM database. The most probable damage pattern is identified through a minimum-error matching criterion, which quantifies discrepancies between the measured and simulated modal responses. This strategy eliminates the need for complex optimization algorithms and provides a rapid, transparent, and physically consistent framework for damage localization and quantification. Moreover, because the HDM is derived from a simplified yet dynamically representative model, the analysis can be executed almost immediately once new data become available, making it highly suitable for rapid post-earthquake emergency response and structural integrity assessments. The HDM-based approach underwent an extensive validation process involving numerical, experimental, and real-world studies to assess its accuracy, robustness, and practical applicability. In the numerical validation phase, the method was applied to a dataset of 1000 numerically modeled shear frame structures featuring diverse stiffness distributions, and damage configurations. This stage verified the mathematical formulation and integration of the HDM terms, ensuring the model's internal consistency and computational stability. Subsequently, the experimental validation involved two independent benchmark studies published in the literature. These experiments provided controlled damage scenarios, allowing a direct comparison between HDM-based damage localization results and the corresponding measured modal periods. The HDM method accurately identified damaged story and stiffness loss, confirming its effectiveness under laboratory conditions. For real-world validation, the HDM algorithm was tested using vibration data from the seven-story Van Nuys Hotel, recorded before and after the 1994 Northridge earthquake. Post-event modal properties were extracted from the acceleration records and analyzed through the HDM framework to estimate the distribution and intensity of structural damage. The resulting damage patterns exhibited a strong correlation with post-earthquake field surveys and visual inspections, reinforcing the credibility of the method. Overall, all three validation stages numerical, experimental, and real-world demonstrated that the HDM-based approach delivers high accuracy, robustness, and computational efficiency. Compared with conventional model updating methods, it offers superior precision with significantly reduced computational cost, making it a promising tool for post-earthquake damage evaluation and real-time structural health monitoring. In addition to the main methodology, the study investigates an analytical correlation between normalized mode shapes and damage sensitivity matrix terms. Preliminary findings suggest that mode shapes may inherently contain sufficient information to infer local stiffness anomalies, suggesting the possibility of developing analytical or MGKhine learning-based predictive models in future research. This discovery could eliminate the need for physical modeling for real-time diagnostics. The broader significance of this research lies in its potential to support smart city applications, where digital twins of critical infrastructures require continuous monitoring and automated decision-making capabilities. The integration of the BSEM and HDM within Internet of Things (IoT)-based sensor networks enables nearinstantaneous evaluation of building integrity following seismic events. Such systems can automatically generate condition reports, prioritize emergency inspections, and guide rescue operations, thereby improving resilience and reducing recovery time after disasters. Another important outcome of this study is its contribution to bridging the gap between academic research and engineering practice. While many existing SHM techniques remain confined to laboratory experiments or small-scale applications, the approach proposed in this thesis was developed with practical implementation in mind. The calibration procedures, data processing techniques, and modeling tools are compatible with widely used engineering software and monitoring systems. Therefore, the proposed framework not only enhances scientific understanding but also offers direct applicability in the field of civil engineering, particularly for structural safety evaluation, retrofitting prioritization, and risk-based maintenance planning. The integration of Simplified Finite Element Modeling (BSEM) with the Damage Sensitivity Matrix (HDM) establishes a new direction in structural diagnostics, where analytical simplicity meets real-world applicability. The results of this research not only validate the scientific foundations of the method but also highlight its transformative potential for post-earthquake safety assessment and long-term structural integrity monitoring. The proposed framework, through its adaptability, efficiency, and clarity, represents a meaningful step toward achieving truly intelligent and autonomous structural health monitoring systems in the near future.

Benzer Tezler

  1. Tez No

    943204

    Çelik-beton kompozit taşıyıcı sistemlerde akışkan viskoz sönümleyici bağlantıların sismik performansa etkisi

    The effect of fluid viscous damper connections on the seismic performance of steel-concrete composite structural systems

    ÖMER FARUK ÖZKAYA

    Yüksek Lisans

    Türkçe

    Türkçe

    2025

    Deprem MühendisliğiErzurum Teknik Üniversitesi

    İnşaat Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ MUHAMMED GÜRBÜZ

  2. Tez No

    1004357

    Kirişli plak ve çift doğrultulu dişli döşemeli bina tasarımlarında plan en boy oranı etkisi

    The effect of plan aspect ratio on building designs with beam-supported slabs and two-way ribbed floors

    EMRE AY

    Yüksek Lisans

    Türkçe

    Türkçe

    2026

    İnşaat Mühendisliğiİstanbul Arel Üniversitesi

    İnşaat Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ MEHDİ ÖZTÜRK

  3. Tez No

    181427

    Betonarme yapı proje hesaplarının STA4-CAD ve Probina Orion paket programlarına göre incelenmesi

    Investigation of reinforced concrete structures projects according to STA4-CAD and Probina Orion programmes

    AYŞEGÜL LÜLE

    Yüksek Lisans

    Türkçe

    Türkçe

    2006

    İnşaat MühendisliğiAfyon Kocatepe Üniversitesi

    Yapı Eğitimi Ana Bilim Dalı

    YRD. DOÇ. DR. ALİ ERGÜN

  4. Tez No

    169932

    Çok katlı betonarme binaların alternatif taşıyıcı sistemlerinin performans ve maliyet açısından incelenmesi

    Performance and cost analysis of multistorey buildings

    MUHAMMET AYDIN ŞERBETÇİOĞLU

    Yüksek Lisans

    Türkçe

    Türkçe

    2005

    İnşaat MühendisliğiKahramanmaraş Sütçü İmam Üniversitesi

    İnşaat Mühendisliği Ana Bilim Dalı

    Y.DOÇ.DR. HATİP TOK

  5. Tez No

    1018221

    Seismic performance of reinforced concrete structural systems under mainshock–aftershock sequences: Insights from the 2023 Türkiye–Syria

    Ana şok–artçı şok dizileri altında betonarme taşıyıcı sistemlerin sismik performansı: 2023 Türkiye–Suriye depremleri ışığında

    VALID ALJOBRANI

    Yüksek Lisans

    İngilizce

    İngilizce

    2026

    Deprem MühendisliğiSakarya Üniversitesi

    İnşaat Mühendisliği Ana Bilim Dalı

    DOÇ. DR. AYDIN DEMİR

Geri Dön