Detail publikace

Prestressed concrete roof girders: Part II – Surrogate modeling and sensitivity analysis

Originální název

Prestressed concrete roof girders: Part II – Surrogate modeling and sensitivity analysis

Anglický název

Prestressed concrete roof girders: Part II – Surrogate modeling and sensitivity analysis

Jazyk

en

Originální abstrakt

The paper describes a particular part of complex stochastic modeling and design of a precast prestressed concrete girder failing in shear, namely surrogate modeling and sensitivity analysis. Both methods were efficiently used in order to reduce high computational effort related to utilization of a 3D nonlinear FEM model. Two types of surrogate models have been developed: (1) artificial neural network model and (2) polynomial chaos expansion model. In case of sensitivity analysis, three methods were utilized and compared: (i) Spearman non-parametric rank-order statistical correlation sensitivity, (ii) sensitivity analysis in terms of coefficient of variation, and (iii) sensitivity analysis in terms of Sobol sensitivity indices. The obtained information was used to set up a stochastic model and surrogate models in an optimum manner and was employed in the subsequent determination of selected uncertain design parameters followed by load-bearing capacity and reliability assessment using semi-probabilistic as well as fully probabilistic approaches.

Anglický abstrakt

The paper describes a particular part of complex stochastic modeling and design of a precast prestressed concrete girder failing in shear, namely surrogate modeling and sensitivity analysis. Both methods were efficiently used in order to reduce high computational effort related to utilization of a 3D nonlinear FEM model. Two types of surrogate models have been developed: (1) artificial neural network model and (2) polynomial chaos expansion model. In case of sensitivity analysis, three methods were utilized and compared: (i) Spearman non-parametric rank-order statistical correlation sensitivity, (ii) sensitivity analysis in terms of coefficient of variation, and (iii) sensitivity analysis in terms of Sobol sensitivity indices. The obtained information was used to set up a stochastic model and surrogate models in an optimum manner and was employed in the subsequent determination of selected uncertain design parameters followed by load-bearing capacity and reliability assessment using semi-probabilistic as well as fully probabilistic approaches.

BibTex


@inproceedings{BUT151810,
  author="David {Lehký} and Drahomír {Novák} and Lukáš {Novák} and Martina {Šomodíková}",
  title="Prestressed concrete roof girders: Part II – Surrogate modeling and sensitivity analysis",
  annote="The paper describes a particular part of complex stochastic modeling and design of a precast prestressed concrete girder failing in shear, namely surrogate modeling and sensitivity analysis. Both methods were efficiently used in order to reduce high computational effort related to utilization of a 3D nonlinear FEM model. Two types of surrogate models have been developed: (1) artificial neural network model and (2) polynomial chaos expansion model. In case of sensitivity analysis, three methods were utilized and compared: (i) Spearman non-parametric rank-order statistical correlation sensitivity, (ii) sensitivity analysis in terms of coefficient of variation, and (iii) sensitivity analysis in terms of Sobol sensitivity indices. The obtained information was used to set up a stochastic model and surrogate models in an optimum manner and was employed in the subsequent determination of selected uncertain design parameters followed by load-bearing capacity and reliability assessment using semi-probabilistic as well as fully probabilistic approaches.",
  address="CRC press, Taylor and Francis group",
  booktitle="Life Cycle Analysis and Assessment in Civil Engineering: Towards an Integrated Vision: Proceedings of the Sixth International Symposium on Life-Cycle Civil Engineering (IALCCE 2018)",
  chapter="151810",
  edition="1",
  howpublished="print",
  institution="CRC press, Taylor and Francis group",
  year="2018",
  month="october",
  pages="495--502",
  publisher="CRC press, Taylor and Francis group",
  type="conference paper"
}