Schliessen
Publizieren
Besondere Sammlungen
Digitalisierungsservice
Hilfe
Impressum
Datenschutz
Schliessen
Publizieren
Besondere Sammlungen
Digitalisierungsservice
Hilfe
Impressum
Datenschutz
zum Inhalt
Detailsuche
Schnellsuche:
OK
Ergebnisliste
Titel
Titel
Inhalt
Inhalt
Seite
Seite
Im Werk suchen
Efficient algorithmic differentiation of CAD frameworks / Mladen Banović ; Advisor: Prof. Dr. Andrea Walther. Paderborn, 2019
Inhalt
Introduction
Motivation
Parametrization approaches: CAD-free vs. CAD-based
Current approaches to evaluate derivatives in CAD environment
Thesis contribution
Other contributions and collaborations
Thesis structure
Test-cases
Parametrization with OCCT
U-bend
TU Berlin stator
TU Berlin stator parametrization in Rolls-Royce workflow
Algorithmic differentiation of OCCT
Introduction to OCCT
Introduction to algorithmic differentiation
Introduction to ADOL-C
Traceless differentiation variant
Trace-based differentiation variant
Approaches of differentiating OCCT
A sample class
The code duplication approach
The inheritance approach
The controller approach
The templating approach
The typedef approach
Compile and run-time issues of OCCT differentiation
Verification of differentiated OCCT
Primal functionality validation of differentiated OCCT
Gradient verification using U-bend parametrization
Gradient verification using TU Berlin stator blade parametrization
Performance tests of differentiated OCCT
Performance of U-bend CAD application
Performance of TU Berlin stator CAD application
Summary
Aerodynamic shape optimization with differentiated OCCT
Mathematical formulation of CAD-based optimization with adjoint method
STAMPS flow solver
Gradient-based shape optimization framework
U-bend optimization results
TU Berlin stator optimization results
Summary
TU Berlin stator optimization with assembly constraints
Implementation of stator assembly constraints
Intersection approach
Interference detection approach based on distance between shapes
Cylinder positioning during shape optimization
Optimization results
Summary
Improved AD of the VKI in-house CAD and grid generation tool
Mesh generation of LS89 axial turbine profile
Structure-exploiting AD of mesh smoothing process
Tailoring trace size to application requirements
Coupling ADOL-C drivers to evaluate derivatives
Handling conditional branches in mesh smoothing
Gradient verification
Performance tests of differentiated CADO
Summary
Algorithmic differentiation of an industrial airfoil design tool
Parablading differentiation
Introduction to Tapenade
Mixed-language AD of Parablading
Parablading differentiation issues
Maintenance aspects
Verification of differentiated Parablading
Geometric derivative validation
Performance test
Gradient-based optimization of TU Berlin stator
Objective functions
CFD setup and optimization workflow
Optimization results
Summary
Conclusion
Future work and research direction
Bibliography
Die detaillierte Suchanfrage erfordert aktiviertes Javascript.