Optimal resource allocation to maximize model accuracy

Dankert, Guillermo Leo

Permalink

https://hdl.handle.net/2142/97467 Copy

Description

Title

Optimal resource allocation to maximize model accuracy

Author(s)

Dankert, Guillermo Leo

Issue Date

2017-04-25

Director of Research (if dissertation) or Advisor (if thesis)

Gardoni, Paolo

Department of Study

Civil & Environmental Eng

Discipline

Civil Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Model uncertainty
• Structural engineering
• Soil structure interaction
• Multi-model uncertainty propagation

Abstract

In the modelling of civil engineering structures, the engineer should make several assumptions to obtain a reasonable prediction of the behavior of the structure. Experience in projects in the professional practice shows that model assumptions made mechanically, without getting into specific details, lead to inaccurate models and to expensive resources allocation that does not render the optimal results. An example within civil engineering is the modelling of a structure, where – typically - structural engineers model the structure above the foundation and geotechnical engineers model the foundation and the soil behavior. These two areas have different error tolerances, mainly because the mechanical uncertainty of manufactured materials as concrete and steel is smaller than the uncertainty of soil. If these differences are not addressed correctly, the optimization process to get more accurate results is done incorrectly. This kind of situation is also common to other interaction with seismic engineering or hydrological engineering, among others. To address the limitations, it is necessary to consider the problem as a system, rather than dealing with the local behavior of the different parts. This work proposes a procedure based on the importance of random and categorical variables to address the different sources of uncertainty. Considering individual component models as categorical variables, the procedure follows a multi-model uncertainty propagation approach. The developed procedure is applied to an example of a cantilever beam on clay foundation.

Graduation Semester

2017-05

Type of Resource

text

Permalink

http://hdl.handle.net/2142/97467

Copyright and License Information

Copyright 2017 Guillermo Dankert

Owning Collections