Upload
fumiya-nozaki
View
959
Download
2
Embed Size (px)
DESCRIPTION
Adjoint法の最大の強みは、設計変数の数に依存しない計算コストで感度を求められる点にあります。例えば、設計変数が2個の場合でも、100個の場合でも、同じ計算コストで各設計変数に関する感度を計算することが可能です。 FRIENDSHIP-Frameworkは、Adjoint法により計算したメッシュの節点ベースの感度分布を通常のCADの寸法パラメータに関する感度に変換する機能を実装することにより、設計変数の数によらないパラメトリック最適化の実現を目指しています。
Citation preview
CAESES / FRIENDSHIP-Framework Users‘ Meeting and Conference 2013
Identifying the Right Parameters Adjoint-parametric sensitivities and visualization
• Usually, a high number of parameters defines a parametric model (about 30-50 for a
typical FRIENDSHIP-Framework model)
• Way too much effort to involve all of them in a conventional optimization process…
How to select the most suitable ones for the optimization task at hand?
• Normally, the designer selects or specifically creates a small number of parameters based
on experience and engineering judgment • This reduces the design space for the optimization
• The designer might not have enough experience to make a good selection
• Especially difficult if the model was created by someone else
Solution: adjoint analysis
Optimization with a Parametric Model
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 2
• Comparison to direct gradient-based method:
J: objective function, α: parameter
What is Adjoint Analysis?
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 3
Adjoint method
Question: what changes to the design
parameters are necessary to get an
optimal product performance?
1 primal + 1 adjoint simulation run for
each objective necessary
Direct method
Question: how do changes of the
design parameters influence the
product‘s performance?
n+1 simulation runs necessary
nn J
J
J
J
33
22
11
optn J ,,,, 321
Result: sensitivity (change in objective function
due to change in parameter value)
• In shape optimization: shape sensitivity (change of
objective function due to normal displacement of
cells on the design boundary)
• A positive shape sensitivity means that the
boundary should be moved in positive normal
direction
• A negative shape sensitivity calls for boundary
movement in negative normal direction
What is Adjoint Analysis?
kn
J
© Volkswagen
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 4
• Sensitivity values can be used to displace the surface cells directly and to morph the
shape, e.g. in a CAD independent approach
• Downside is that the shape changes cannot easily be fed back into the design workflow
Solution: connect shape sensitivities to CAD parameters
• Parametric sensitivity
How to use the Adjoint Sensitivities
n
k
kn
n
n
JJ
Adjoint shape
sensitivity
Normal displacement of model boundary due to
CAD parameter changes: „design velocity“
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 5
• Normal displacement of the boundary due to change of CAD parameter value
• Determine by perturbing the parameters and measuring the normal displacement of given
positions on the model boundary (tessellation nodes of surfaces or trimeshes)
Sensitivity Computation
• Input: model boundary as surfaces or trimeshes
• Automatically determines all design variables that
are suppliers to the boundary
• User can deselect design variables and set
individual deltas
• Result: design velocity maps for all selected
parameters
Design Velocity
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 6
• Also computed by Sensitivity Computation
• Adjoint sensitivities are mapped to the positions on the boundary used to determine the
normal displacement (tessellation nodes) by probing the result data set of the adjoint
computation
• Adjoint sensitivities are multiplied with the design velocities for each parameter
• Result: one scalar parametric sensitivity for each parameter
Parametric Sensitivities
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 7
• Parameter with small influence
Parametric Sensitivities
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 8
• Parameter with large influence
Parametric Sensitivities
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 9
• From the list of parametric sensitivities select the parameters with the biggest influence on
the objective function follow up with conventional optimization
or
• Multiply the vector of the parametric sensitivities with a step size factor and add to the
parameter values determine the right step size with 1-dimensional search using primal
CFD computations
or
• …?
Parametric Sensitivities
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 10
• Largest possible design space can be considered
• All parameters in the model can be used for optimization, without the need for the user to
understand the effect of each parameter
• The effort does not scale with the number of parameters
• Parametric sensitivities for all parameters in the model can be computer more quickly than
for a small set using the direct approach
But:
• Predictions based on sensitivities are only valid for small changes in shape
• In practice iterative process: new sensitivities should be computed for modified shape
Also:
• Apart from the connection to adjoint analysis, the sensitivity computation can be used to
understand the effect of the parameters on the model
Discussion
Users‘ Meeting and Conference 2013 | Best Practices and Anniversary | 11