Probabilistic structural integrity assessment based on direct methods

  • Xiaoxiao WANG

Student thesis: Doctoral Thesis


Structural integrity assessment is an effective way to measure the safety of criticalinfrastructures under the complicated combination of high-temperature and varying load conditions. However, under the urgent need for carbon peaking and carbon neutrality, the current industry has to pursue extreme operating parameters and precise design solutions, which inevitably involve a multiplicity of uncertainties in design considerations. Unfortunately, the majority of basic evaluation procedures are dependent on deterministic analysis approaches, with expert experience based-safety factors accounting for the randomness. Due to the lack of statistical characterization of key parameters for failure analysis, this scheme tends to cause conservativeness and offset the benefits gained from the development of advanced computational methods. Therefore, it is crucial to develop a plausible probabilistic structural integrity assessment framework in terms of computational efficiency and accuracy. This thesis reviews the latest research progress on the structural integrity assessment for high temperature structures and delivers a new insight into the probabilistic structural integrity assessment framework based on the direct method and artificial intelligence technology. Firstly, by systematically comparing three different creep rupture analysis methods, a quasi-efficient deterministic analysis method for high-temperature structure is identified for the subsequent probabilistic structural integrity assessment framework. Secondly, the cyclic plastic response of the cracked specimen is investigated by Linear Matching Method (LMM) considering the crack constraint effect on the alternating plasticity and ratchet limit, where the capability of the selected numerical method to deal with the structures in the presence of the defect is demonstrated in detail. Thirdly, aiming at predicting the structural failure probability of violating shakedown condition, the probabilistic shakedown analysis under the Linear Matching Method (pLMM) framework is proposed based on the LMM shakedown procedure and First Order Reliability Method (FORM). Furthermore, taking advantage of the Artificial Neural Network (ANN) technique, the probabilistic Low Cycle Fatigue (LCF), ratcheting and creep-fatigue analyses are also established, with the physics-based surrogate model constructed and trained by LMM-driven dataset. The key designparameters that influence the structural ratcheting limit, LCF life and creep-fatigue life are revealed and discussed in depth, and the probabilistic assessment curves for engineering components are built in terms of ratcheting, LCF and creep-fatigue failure modes, with the reliability-based safety factors calibrated considering multi-reliability requirements. This study is dedicated to the probabilistic structural integrity assessment strategies covering extensive failure mechanisms and conducive to achieving better reliability-centred risk management for critical infrastructures.
Date of Award5 Jun 2023
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde
SupervisorHaofeng Chen (Supervisor) & David Nash (Supervisor)

Cite this