Slawomir Koziel
Reykjavik University
Iceland
e-mail

Leifur Leifsson
Reykjavik University
Iceland
e-mail The use of computer simulations is ubiquitous in contemporary engineering design. High-fidelity numerical models are very accurate but, at the same time, computationally expensive and inherently noisy. Depending on the structure complexity and required accuracy level, evaluation time for a single design can take hours, days or even weeks. Therefore, direct use of high-fidelity simulations in the optimization loop may be prohibitive. On the other hand, in many cases, simulation-driven design is the only option due to complexity of the structure under consideration and the lack of analytical models and/or systematic design procedures. In such instances, computationally efficient design can be performed using surrogate-based optimization (SBO), where the high-fidelity model is replaced by its computationally cheap but still reasonably accurate representation, a surrogate. The surrogate model can be created using various approximation schemes. The surrogate can also be knowledge-based, i.e., constructed from the low-fidelity model that enjoys the same physics as the high-fidelity one. One of the important goals of SBO procedures is to reduce the number of high-fidelity model evaluations, and, consequently, to lower the overall optimization cost. The special session will focus on reviewing state-of-the-art and promoting new directions of surrogate-based and knowledge-based design optimization and modeling methodologies exploiting computer simulations as well as their applications in various fields of science and engineering. Topics of interest include, but are not limited to:

• Computationally Efficient Optimization of Expensive Objective Functions
• Simulation-driven Design
• Function-approximation-based and Physics-based Surrogate Models
• Surrogate-based Modeling and Optimization
• Multi-fidelity and Variable-resolution Analysis and Optimization
• Knowledge-based Methods
• Response Surface Approximation, Space Mapping, and Response Correction Techniques
• Multidisciplinary Design and Optimization
• Adjoint Sensitivities in Simulation-driven Design
• Software Architectures for Optimization-oriented Design
• Application Case Studies

All accepted papers will be published in a special section of the proceedings book, under an ISBN reference and on CD-ROM support.
All papers presented at the conference venue will be available at the SciTePress Digital Library (http://www.scitepress.org/DigitalLibrary/). SciTePress is member of CrossRef (http://www.crossref.org/).

 

Last update: 22.9.2011 · © INSTICC