07.02.2013 Views

Optimization and Computational Fluid Dynamics - Department of ...

Optimization and Computational Fluid Dynamics - Department of ...

Optimization and Computational Fluid Dynamics - Department of ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

5 Efficient Deterministic Approaches for Aerodynamic Shape <strong>Optimization</strong> 121<br />

<strong>and</strong><br />

δxξ,...,δyη =0,δw=0 on B = B(X) (5.21)<br />

are also solved just once.<br />

The right h<strong>and</strong> side −d(I) <strong>of</strong> the wall boundary condition <strong>of</strong> the quasiunsteady<br />

adjoint Euler equations is dependent on the cost function I. The<br />

adjoint far-field boundary condition just states that the geometrical position<br />

<strong>of</strong> the far-field is fixed <strong>and</strong> free stream conditions apply there.<br />

Finally, the components <strong>of</strong> the gradient ∇XI =(δiI)i=1,...,n can now be<br />

determined via an integration just over the adjoint solution <strong>and</strong> the metric<br />

sensitivities δxξ,...,δyη <strong>and</strong><br />

�<br />

δI = − p(−ψ2δyξ + ψ3δxξ) dl + K(I)<br />

C<br />

�<br />

−<br />

ψ<br />

D<br />

⊤ ξ (δyηf − δxηg)+ψ ⊤ η (−δyξf + δxξg) dA (5.22)<br />

is obtained where K(I) is again a term dependent on the cost function I.<br />

For the gradient <strong>of</strong> the drag, the following right h<strong>and</strong> side adjoint boundary<br />

on C is used<br />

d(CD)=<br />

2<br />

γM 2 ∞ p∞Cref<br />

(nx cosα + ny sinα) (5.23)<br />

<strong>and</strong> to get the corresponding gradient, K(I) is<br />

K(CD)= 1<br />

�<br />

Cp(δnx cosα + δny sinα) dl (5.24)<br />

for the gradient <strong>of</strong> the lift<br />

<strong>and</strong><br />

d(CL)=<br />

Cref<br />

K(CL)= 1<br />

C<br />

2<br />

γM 2 ∞ p∞Cref<br />

Cref<br />

are used, <strong>and</strong> for the gradient <strong>of</strong> the pitching moment<br />

<strong>and</strong><br />

d(Cm)=<br />

2<br />

�<br />

C<br />

γM 2 ∞ p∞C 2 ref<br />

(ny cosα − nx sinα) (5.25)<br />

Cp(δny cosα − δnx sinα) dl (5.26)<br />

(ny(x − xm) − nx(y − ym)) (5.27)<br />

K(Cm)= 1<br />

C2 �<br />

Cpδ(ny(x − xm) − nx(y − ym)) dl (5.28)<br />

ref C<br />

are used. For more details see [7] or [8].

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!