12.01.2013 Views

Design Optimization of Flow Path with ANSYS-Workbench and ...

Design Optimization of Flow Path with ANSYS-Workbench and ...

Design Optimization of Flow Path with ANSYS-Workbench and ...

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.

<strong>Design</strong> <strong>Optimization</strong> <strong>of</strong> <strong>Flow</strong> <strong>Path</strong> <strong>with</strong><br />

<strong>ANSYS</strong>-<strong>Workbench</strong> <strong>and</strong> optiSLang<br />

Johannes Einzinger<br />

<strong>ANSYS</strong> Continental Europe<br />

johannes.einzinger@ansys.com<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 1 <strong>ANSYS</strong>, Inc. Proprietary


Outline<br />

• Motivation<br />

• Preliminary Consideration<br />

• Practical Example: <strong>Flow</strong> through Air Condition<br />

• Practical Example: Gas Turbine Stage<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 2 <strong>ANSYS</strong>, Inc. Proprietary


Motivation<br />

Power Plant<br />

Efficiency<br />

Increase <strong>of</strong> 1%<br />

=Electricity for<br />

1000 MW<br />

50 %<br />

+20 MW<br />

120 000<br />

Inhabitants<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 3 <strong>ANSYS</strong>, Inc. Proprietary


Preliminary Consideration<br />

optiSLang<br />

Parametric<br />

Geometry<br />

<strong>ANSYS</strong> <strong>Workbench</strong><br />

Automatic<br />

Meshing<br />

Automatic<br />

CFD-Solution<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 4 <strong>ANSYS</strong>, Inc. Proprietary


Preliminary Consideration<br />

CFD Requirements<br />

• Complex geometries<br />

• High dem<strong>and</strong> on meshing (Boundary Layer…)<br />

• Relatively long computation time<br />

Best Practice<br />

• Error estimation for CFD Model<br />

• Model reduction (simple to complex model)<br />

• <strong>Optimization</strong> Strategy<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 5 <strong>ANSYS</strong>, Inc. Proprietary


Practical Example:<br />

<strong>Flow</strong> through Air Condition<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 6 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Flow</strong> through Air Condition<br />

• Geometry Parameterization<br />

• Parametric Meshing<br />

• CFD-Simulation Set-Up<br />

• <strong>Optimization</strong><br />

– <strong>Design</strong> <strong>of</strong> Experiments (DoE)<br />

– Adaptive Response Surface Method (ARSM)<br />

– Evolutionary Algorithm (EA)<br />

– Pareto <strong>Optimization</strong> (Pareto)<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 7 <strong>ANSYS</strong>, Inc. Proprietary


Geometry Parameterization<br />

Deflector device made <strong>of</strong><br />

five identical blades<br />

α<br />

Heat Source<br />

Radius unten<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 8 <strong>ANSYS</strong>, Inc. Proprietary<br />

α<br />

L4<br />

L3<br />

R1<br />

R2<br />

Anstellwinkel<br />

Dicke vorn<br />

Dicke hinten<br />

Radius oben


Parametric Meshing<br />

• Hex-mesh for<br />

(1), (2), (3),<br />

static volumes<br />

• Tet-Prism-mesh<br />

for flexible<br />

volume<br />

• Tets in the<br />

volume (4)<br />

• Prism for<br />

boundary layer<br />

resolution (5)<br />

(1)<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 9 <strong>ANSYS</strong>, Inc. Proprietary<br />

(5)<br />

(4)<br />

(2)<br />

(3)


CFD-Simulation Set-Up<br />

Inlet: 10 [m/s], 300 [K]<br />

P 1 Heat source (50 W)<br />

Medium: Air<br />

Lamellenkraft<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 10 <strong>ANSYS</strong>, Inc. Proprietary<br />

T 1<br />

c 1<br />

p v<br />

F L<br />

T 1<br />

p v<br />

Output<br />

Temperatur P 1<br />

Geschw. P 1<br />

Druckverlust<br />

Objective<br />

= T Target<br />

=min


<strong>Design</strong> <strong>of</strong> Experiments<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 11 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Design</strong> <strong>of</strong> Experiments<br />

Lammelenkraft Druckverlust<br />

Anstellwinkel<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 12 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Design</strong> <strong>of</strong> Experiments<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 13 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Optimization</strong><br />

Initial <strong>Design</strong><br />

Best <strong>of</strong> DoE<br />

ARSM<br />

EA<br />

Pareto<br />

Input Parameter:<br />

Winkel α<br />

Radius oben R O<br />

Dicke hinten D H<br />

-13.0<br />

306.6<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 14 <strong>ANSYS</strong>, Inc. Proprietary<br />

α [°]<br />

0.0<br />

-23.7<br />

-25.0<br />

-16.1<br />

R O<br />

[mm]<br />

60.0<br />

59.8<br />

63.2<br />

49.4<br />

42.0<br />

D H<br />

[mm]<br />

8.0<br />

6.9<br />

6.8<br />

6.0<br />

8.0<br />

Target:<br />

Min(T 1 -T Target )<br />

+Min(p V )<br />

T 1<br />

[K]<br />

326.0<br />

304.0<br />

304.8<br />

304.0<br />

p V<br />

[Pa]<br />

14.0<br />

49.7<br />

63.0<br />

27.7<br />

45.6


<strong>Optimization</strong><br />

DoE<br />

EA<br />

Pareto<br />

ARSM<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 15 <strong>ANSYS</strong>, Inc. Proprietary


Summary<br />

• <strong>Design</strong> <strong>of</strong> Experiment<br />

– Shows Pareto Front (100 <strong>Design</strong>s)<br />

• Adaptive Response Surface Method<br />

– Finds Point on Pareto Front (71 <strong>Design</strong>s)<br />

• Evolutionary Algorithm<br />

– Finds Point on Pareto Front (105 <strong>Design</strong>s)<br />

• Pareto <strong>Optimization</strong><br />

– Finds best point (108 <strong>Design</strong>s)<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 16 <strong>ANSYS</strong>, Inc. Proprietary


Practical Example:<br />

Gas Turbine Stage<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 17 <strong>ANSYS</strong>, Inc. Proprietary


Gas Turbine Stage<br />

• Set-Up <strong>and</strong> Parameterization<br />

• <strong>Optimization</strong><br />

• Strategy 1:<br />

– <strong>Design</strong> <strong>of</strong> Experiments (DoE)<br />

– Adaptive Response Surface Method (ARSM),<br />

based on good DoE <strong>Design</strong>(s)<br />

• Strategy 2:<br />

– Evolutionary Algorithm (EA), global search<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 18 <strong>ANSYS</strong>, Inc. Proprietary


Set-Up <strong>and</strong> Parameterization<br />

Boundary Conditions<br />

Total Pressure@Inlet<br />

Total Temperature@Inlet<br />

Mass <strong>Flow</strong> Rate@Outlet<br />

Rotational Velocity<br />

Per Segment<br />

0.25 [atm]<br />

340 [K]<br />

0.06 [kg/s]<br />

5000 [rev/min]<br />

Output Parameter<br />

Total Pressure Ratio<br />

Total Temperature Ratio<br />

Isentropic Efficiency<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 19 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Design</strong> <strong>of</strong> Experiments<br />

Total Temperature Ratio<br />

Total Pressure Ratio<br />

Isentropic Efficiency<br />

Mass <strong>Flow</strong> Rate<br />

Rotational Velocity<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 20 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Design</strong> <strong>of</strong> Experiments<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 21 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Design</strong> <strong>of</strong> Experiments<br />

Total Pressure Ratio<br />

Mass <strong>Flow</strong> Rate<br />

Isentropic Efficiency<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 22 <strong>ANSYS</strong>, Inc. Proprietary


<strong>Optimization</strong><br />

ARSM<br />

EA<br />

Input Parameter:<br />

Mass <strong>Flow</strong> Rate<br />

Rotational Velocity<br />

Initial <strong>Design</strong><br />

Best <strong>of</strong> DoE<br />

Mass<br />

<strong>Flow</strong> Rate<br />

[g/s]<br />

60.0<br />

61.2<br />

60.7<br />

60.4<br />

Target:<br />

Isentropic Efficiency = MAX<br />

Rotational<br />

Velocity<br />

[U/min]<br />

5000<br />

4800<br />

4750<br />

4550<br />

Isotropic<br />

Efficiency<br />

[%]<br />

87.5<br />

87.9<br />

88.0<br />

88.1<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 23 <strong>ANSYS</strong>, Inc. Proprietary


Pressure Distribution<br />

Initial <strong>Design</strong> <strong>Design</strong> <strong>of</strong> Experiments<br />

Adaptive Response Surface Evolutionary Algorithm<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 24 <strong>ANSYS</strong>, Inc. Proprietary


Mach Number, Blade to Blade Plot<br />

Initial <strong>Design</strong> <strong>Design</strong> <strong>of</strong> Experiments<br />

Adaptive Response Surface<br />

Evolutionary Algorithm<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 25 <strong>ANSYS</strong>, Inc. Proprietary


Summary<br />

• Improvement <strong>of</strong> Isentropic Efficiency +0.5%<br />

• Axial Turbine has a global maximum <strong>of</strong> Efficiency<br />

• Strategy 1 (DoE+ARSM):<br />

– Number <strong>of</strong> <strong>Design</strong>s 53 (DoE=22, ARSM=31)<br />

– Efficient to find “the maximum”<br />

• Strategy 2 (EA):<br />

– Number <strong>of</strong> <strong>Design</strong>s 105<br />

– Finds “the maximum” in parameter space<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 26 <strong>ANSYS</strong>, Inc. Proprietary


Thank You!<br />

© 2008 <strong>ANSYS</strong>, Inc. All rights reserved. 27 <strong>ANSYS</strong>, Inc. Proprietary

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

Saved successfully!

Ooh no, something went wrong!