Thursday, November 3, 2016

Call for papers!

Call for papers!! Special Issue: "Evolutionary Algorithms and Metaheuristics: Applications in Engineering Design and Optimization". Mathematical Problems in Engineering Journal. +info: https://www.hindawi.com/journals/mpe/si/432036/cfp/

Wednesday, February 24, 2016

EUROGEN 2017

INTERNATIONAL CONFERENCE ON EVOLUTIONARY AND DETERMINISTIC METHODS FOR DESIGN OPTIMIZATION AND CONTROL WITH APPLICATIONS TO INDUSTRIAL AND SOCIETAL PROBLEMS

Sep 13-15, 2017 in Madrid, Spain



Topics:

  • Metaheuristics and Evolutionary Algorithms (including Evolutionary Programming, Evolution Strategies, Genetic Algorithms, Memetic Algorithms and Artificial Immune Systems, etc.)
  • Multi-objective Evolutionary Algorithms and Constraint Handling Techniques
  • Adjoint based and One-shot Methods
  • Hybrid Optimisation methods (Gradient Based Methods, Combinatorial Optimization Methods, etc.)
  • High Performance Computing and GPUs based Optimisation Algorithms
  • Goal oriented optimization for mesh and meshless methods
  • Game Strategies
  • Surrogate Models for Optimisation
  • Parallel and distributed Evolutionary Algorithms (from LANs to GRID)
  • Multi-disciplinary Optimization Methods
  • Design Optimisation under Uncertainties
  • Multi-criteria Decision Making
  • Topology Optimization

with applications to the following areas:

  • Aeronautics, Aerospace, and Space
  • Agriculture and Food Production
  • Automotive
  • Bio Engineering and Medicine
  • CAD and Manufacturing
  • Civil and Structural Engineering
  • Economics and Finance
  • Electromagnetics
  • Electronics
  • Energy, including Renewable Energy
  • Environment
  • Genetics
  • Material Science and Technology
  • Offshores, Coastal and Marine Engineering and Operations
  • RAMS Analysis
  • Risk Analysis
  • Smart Materials and Structures
  • Social Sciences and Politics
  • Telecommunications
  • Transport and Logistics
  • Turbomachinery
Congress web: http://eurogen2017.etsiae.upm.es/

Tuesday, October 20, 2015

Mini-symposium on "Surrogate-assisted Evolutionary Algorithms in Aerodynamic Design/Optimization" at ECCOMAS 2016


During the last decades evolutionary algorithms (EAs) have become an everyday tool in aerodynamic design/optimization problems. EAs can handle complex, constrained, multi-objective optimization problems with noisy objective functions using any analysis software as "black-box".
However, they require a vast number of evaluations to reach the optimal solutions even in problems with a few design variables. Since, in aerodynamic optimization, each evaluation involves a CFD, CSM, etc analysis, EAs become computationally prohibitive for real-world applications.
Therefore, there has been a raising interest in techniques that may reduce the computational cost of an EA-based optimization.  The most popular among these is the use of surrogate evaluation models (or metamodels).
Metamodels are low cost approximations to the costly analysis model (CFD or CSM software etc.) able to provide a fast and sufficiently accurate estimation of the quantities of interest of the simulation (e.g., the aerodynamic coefficients) and, hence, of the objective function used in the optimization process.
This minisymposium aims at collecting and disseminating new ideas in surrogate modeling and surrogate-based optimization for aerodynamic design.  Emphasis is laid on the development and use of fast and efficient metamodels for CFD-based real-world industrial optimization problems.



Organized by:

Varvara Asouti (National Technical University of Athens, Greece)
Esther Andrés (Spanish National Institute for Aerospace Technology, Spain)
Emiliano Iuliano (Centro Italiano Ricerche Aerospaziali, Italy)


More info.:http://www.eccomas2016.org/

Monday, August 17, 2015

EUROGEN 2015 programme



The EUROGEN 2015 preliminary programme has been published: http://www.strath.ac.uk/eurogen2015/programme/

The draft programme of the mini-symposium on "MS1 - Surrogate-Based Optimization in Aerodynamic Design" is:
  • Constrained Single-Point Aerodynamic Shape Optimization of the DPW-W1 wing through Evolutionary Programming and Support Vector Machines, Esther Andrés, Daniel González-Juárez, Mario J. Martin-Burgos and Leopoldo Carro-Calvo - Monday 14th, 10:20 to 10:40, room A, Level 3
  • Efficient Global Optimization method for multipoint airfoil design, Davide Cinquegrana and Emiliano Iuliano - Monday 14th, 10:40 to 11:00, room A, Level 3
  • Artificial Neural Networks for Surrogate-based Optimization in Preliminary Aerodynamic Design, Petr Dvorák - Monday 14th, 11:00 to 11:20, room A, Level 3
  • Application of Surrogate-based Optimization Techniques to Aerodynamic Design Case, Emiliano Iuliano and Domenico Quagliarella -Monday 14th, 11:20 to 11:40, room A, Level 3
  • A Review of Surrogate Modeling Techniques for Aerodynamic Analysis and Optimization: Current Limitations and Future Challenges in Industry, Raul Yondo Mine, Kamil-Krzysztof Bobrowski, Esther Andrés and Eusebio Valero - Monday 14th, 11:40 to 12:00, room A, Level 3
  • Use of Surrogate Models for the Global Optimization of the Shapes of Flying Configurations, in Supersonic Flow, Adriana Nastase - Monday 14th, 12:00 to 12:20, room A, Level 3
  • Enabling of Large Scale Aerodynamic Shape Optimization through POD-based Reduced-Order Modeling and Free Form Deformation, Angela Scardigli, Rocco Arpa and Haysam Telib - Monday 14th, 12:20 to 12:40, room A, Level 3

Thursday, February 5, 2015

Minisymposium accepted for the EUROGEN 2015 conference



International Conference on Evolutionary and Deterministic Methods for Design,
Optimization and Control with Applications to Industrial and Societal Problems
EUROGEN 2015
Glasgow, UK, 14-16 September 2015

Mini-symposium on Surrogate-based optimization in aerodynamic design

Key words: Aerodynamic shape design, Surrogate Modeling, Global optimization methods

ABSTRACT
Aerodynamic design, with particular reference to its early stages, requires exploring the design space in a global sense in order to locate the optimal candidate. Global optimization methods (e.g., evolutionary algorithms) can meet this requirement as they have the ability to work with noisy objective functions without assumptions on continuity and with a high potential to find the optimum of complex problems. However, they involve a vast number of evaluations even for a small number of design variables. As each evaluation requires a CFD complete analysis, this would make the method unfeasible, in terms of computational cost, for real-world applications. Therefore, there has been a raising interest in surrogate modelling which promises to provide sufficiently accurate solution of complex problems with limited or reduced computational efforts. Surrogate modelling methods are widely recognized as supporting tools easing the designer task in design space exploration, parametric studies, visualization and optimization. With the term “surrogate” it is usually meant an analysis method which, in some sense, is “alternative” to the high-fidelity one (e.g., CFD) and, with respect to it, is able to provide a quicker and sufficiently accurate estimation of the quantities of interest of the simulation (e.g., the aerodynamic coefficients) and, hence, of the fitness function in an optimization process. Recently, a GARTEUR Action Group has been established to explore surrogate-based global optimization approaches. The main objective of the AG work [1] is, by means of a European collaborative research, to make a deep evaluation and assessment of surrogate-based global optimization methods for aerodynamic shape optimization, dealing with the main challenges as the curse of dimensionality, reduction of the design space and error metrics for model validation, amongst others. The special session will provide a chance to share recent progresses and results in the field of surrogate-based aerodynamic optimization, to meet up with other researchers in the related field and to exchange ideas on the topic in a fruitful discussion.

TOPICS
The special session is aimed at collecting and disseminating contributions and new ideas in surrogate modelling and surrogate-based optimization for aerodynamic design. A particular emphasis is put on the development of fast and efficient meta-models for CFD-based optimization applications where multimodality, high non-linearity, non-differentiability, high dimensionality and high computational cost are expected features. For these reasons, real-world industrial applications are welcome.
Authors are invited to submit papers on one or more of the following topics:

  • Data fitting surrogate modeling
  • Multi-fidelity surrogate modeling
  • Global and Local Surrogate-based optimization (SBO)
  • Efficient Global Optimization
  • Adaptive sampling strategies
  • Trust region approaches
  • Proper Orthogonal Decomposition methods
  • Reduced Order modeling
  • Evolutionary search methods

REFERENCES

[1] GARTEUR EG67 members. Proposal for the establishment of a GARTEUR Action Group: “Surrogate-based global optimization methods in preliminary aerodynamic design”, November 2012.

IMPORTANT DATES:

Thursday, July 31, 2014

Minisymposim on "Surrogate-based global optimization methods for aerodynamic shape design" at ECCOMAS CFD 2014

The GARTEUR AG52 organized a Minisymposim on "Surrogate-based global optimization methods for aerodynamic shape design" during the international conference ECCOMAS CFD 2014, which took place in Barcelona 20-25 July.

Thank you very much to all the speakers!!


Monday, May 26, 2014

CFD cross-analysis preliminary results

In order to identify the discrepancies produced by the use of different CFD solvers, a CFD cross-analysis is performed on the baseline RAE2822 and DPW configurations. The following tables shows the partners’ results for both defined design points and the Cp plots comparison for the RAE2822 test case.
RAE2822 CFD cross-analysis results

DP1
CL=0.8, Mach=0.734, Re=6.5*106
SA turbulence model
DP2
CL=0.743, Mach=0.754, Re=6.2*106
SST turbulence model

AoA
Cl
Cd
Cmy
AoA
Cl
Cd
Cmy
AG-138 Ref.
3.19
0.8030
0.0168
-0.0990
3.19
0.7430
0.0242
-0.1060
CIRA / ZEN
2.65
0.8025
0.0189
-0.1008
2.90
0.7431
0.0286
-0.1009
INTA / TAU
2.79
0.8005
0.0188
-0.0953
2.12
0.7435
0.0240
-0.1215
ONERA / ELSA
2.72
0.8000
0.0194
-0.1011
2.65
0.7424
0.0276
-0.1101
UNIS / VGK
2.70
0.8000
0.0173
-0.0921
2.47
0.7428
0.0215
-0.1031
VUT / EDGE
2.89
0.8000
0.0174
-0.0915
3.00
0.7430
0.0263
-0.0971
Max. difference
-
-
0.0021
0.0096
-
-
0.0071
0.0244
% (with respect to mean)
-
-
11%
9%
-
-
27%
22%

The differences encountered in DP1 are within reasonable margins. However DP2 shows a higher sensitivity with respect to the CFD solver used by each partner. In addition, for this design point, some partners found difficulties to achieve good convergence (at least reasonable for optimization purposes) using a Menter SST turbulence model. In particular, INTA values are computed using a Wilcox-k-w+SST turbulence model instead of the Menter one, as it was not possible to get a good convergence with that model. This also explains why the observed differences are higher than in DP1. In any case, the optimized geometries will be also cross-validated and after Task 1.2, the returned optimized geometries will be evaluated using a unique CFD solver.

Cp distribution plots

A CFD cross-analysis has also been performed on the baseline DPW wing. The following table shows the participating partners’ results and next figure displays the Cp plots comparison.

DPW Euler CFD cross-analysis results for the given flow conditions

Mach=0.76, AoA=0.5
Cl
Cd
Cmy
INTA / TAU
0.6077
0.0208
-0.1037
SAAB /EDGE
0.6072
0.0210
-0.1044
FOI / EDGE
0.6071
0.0210
-0.1044
ONERA / ELSA
0.5920
0.0185
-0.0989
Max. Difference
0.0157
0.0025
0.0055
% (with respect to mean)
2%
12%
4%
         
In this test case, ONERA results are the only ones using a structured grid/solver.