Reliability Analysis And Safety Assessments of Structural Wall with Nonlinear Finite Element Analyses

Abstract

Inclusion of safety formats method in nonlinear finite element analysis of a structure is beneficial to realize
the safe and real behavior of structures. In addition, reliability analysis in combination with NLFEA results
and safety assessments provide an insight into consistency of the safety assessments recommendations related
to targeted safety. However, various complications arise regarding the appropriate constitutive model
to be used in analysis, estimation and inclusion of modelling uncertainty in results from nonlinear finite element
analysis, unavailability of a suitable function that defines the response of a structure when subjected to
a certain load etc.
In this additional thesis, a case study is performed to study the combination of structural safety assessment
and reliability analysis using nonlinear finite element analysis. A wall specimen experimented by Lefas
et. al is selected for the study [1]. Solution strategy is adopted based on the recommendations ofDutch guidelines
with some deviations [2]. The nonlinear finite element analyses are performed inDIANA 10.1. Since only
one solution strategy was used during the analysis of a single wall specimen, the modelling uncertainty parameters
were estimated using results from previous studies[3]. The modelling uncertainty parameters were
estimated to be, µm Æ 1.21 and Vµ Æ 6.88%. These parameters infer that the results obained from different
solution strategies were close to each other but still had a considerable bias with respect to the structural
capacity obtained from the experimental values. The NLFEA underestimated the capacity as compared to
experimental results. Only 79.3 % of the experimental capacity was realized from NLFEA.
Three different safety format methods suggested in [2] and a new safety format method by Schlune et al.
[4] are used in NLFEA to estimate the design load capacity. The effect of inclusion of estimated model uncertainty
parameters on design load capacity is observed too. The results of safety assessments indicated that
the design load estimated by using ECOV and Schlune et al. safety format are comparatively higher than using
Partial Safety Factor and Global Resistance Factor, but still conservative with respect to the experimental
results.
The reliability of the design load estimated by using different safety assessment methods is checked using
response surface method and first order reliability methods. A response surface method together with FORM
is used to obtain a quadratic limit state function that closely represents the response of the structural wall
subjected to the design load value. The limit state function thus obtained is then used for FORM analysis,
Importance sampling and/or Monte Carlo simulations to observe if the intended safety level is achieved.
The study of acheived reliability index indicated that the design load values estimated by using three safety
formats recommended in [2] are conservative, whereas the one estimated by using Schlune et al. format was
found to be non-conservative.
The additional thesis was carried out with various approximations and limitations. All limitations and approximations
are explained in detail and justified where necessary and possible. The results and discussions
of this additional thesis are dependent only on a single solution strategy adopted in a single wall specimen
and in a single failure mode. Therefore, the results should instead be interpreted as an indication of need of
further and extensive research and studies required in the field of reliability analysis and safety assessment
using nonlinear finite element analysis.