design optimization of flow path with ansys …© 2008 ansys, inc. all rights reserved. 1 ansys,...

© 2008 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary

Design Optimization of Flow Path with ANSYS-Workbench and optiSLang

Design Optimization of Flow Path with ANSYS-Workbench and optiSLang

Johannes EinzingerANSYS Continental Europe

[email protected]

Johannes EinzingerANSYS Continental Europe

[email protected]


Outline

• Motivation

• Preliminary Consideration

• Practical Example: Flow through Air Condition

• Practical Example: Gas Turbine Stage


Motivation

120 000

Inhabitants

=Electricity for

+20 MWIncrease of 1%

50 %Efficiency

1000 MWPower Plant


Preliminary Consideration

ParametricGeometry

AutomaticMeshing

AutomaticCFD-Solution

ANSYS Workbench

optiSLang


Preliminary Consideration

CFD Requirements

• Complex geometries

• High demand on meshing (Boundary Layer…)

• Relatively long computation time

Best Practice

• Error estimation for CFD Model

• Model reduction (simple to complex model)

• Optimization Strategy


Practical Example:Flow through Air Condition

Practical Example:Flow through Air Condition


Flow through Air Condition

• Geometry Parameterization

• Parametric Meshing

• CFD-Simulation Set-Up

• Optimization

– Design of Experiments (DoE)

– Adaptive Response Surface Method (ARSM)

– Evolutionary Algorithm (EA)

– Pareto Optimization (Pareto)


Geometry Parameterization

Radius untenR2

Radius obenR1

Dicke hintenL3

Dicke vornL4

Anstellwinkelα

Heat Source

Deflector device made of five identical blades

α


Parametric Meshing

• Hex-mesh for (1), (2), (3), static volumes

• Tet-Prism-mesh for flexible volume

• Tets in the volume (4)

• Prism for boundary layer resolution (5)

(4)

(5)

(2)

(1)

(3)


CFD-Simulation Set-Up

P1 Heat source (50 W)

Inlet: 10 [m/s], 300 [K]

Medium: Air LamellenkraftFL

Druckverlustpv

Geschw. P1c1

Temperatur P1T1

Output

=minpv

= TTargetT1

Objective


Design of Experiments



Lammelenkraft Druckverlust

Anstellwinkel


Optimization

27.7306.66.049.4-13.0EA

45.6304.08.042.0-16.1Pareto

6.8

6.9

8.0

DH

[mm]

304.8

304.0

326.0

T1

[K]

pV

[Pa]RO

[mm]α [°]

63.063.2-25.0ARSM

49.759.8-23.7Best of DoE

14.060.00.0Initial Design

Input Parameter:Winkel α

Radius oben RO

Dicke hinten DH

Target:Min(T1-TTarget)

+Min(pV)


Optimization

DoE

ARSMEA

Pareto


Summary

• Design of Experiment

– Shows Pareto Front (100 Designs)

• Adaptive Response Surface Method

– Finds Point on Pareto Front (71 Designs)

• Evolutionary Algorithm

– Finds Point on Pareto Front (105 Designs)

• Pareto Optimization

– Finds best point (108 Designs)


Practical Example:Gas Turbine StagePractical Example:Gas Turbine Stage


Gas Turbine Stage

• Set-Up and Parameterization

• Optimization

• Strategy 1:

– Design of Experiments (DoE)

– Adaptive Response Surface Method (ARSM),

based on good DoE Design(s)

• Strategy 2:

– Evolutionary Algorithm (EA), global search


Set-Up and Parameterization

5000 [rev/min]Rotational Velocity

0.06 [kg/s]Mass Flow Rate@Outlet

340 [K]Total Temperature@Inlet

0.25 [atm]Total Pressure@Inlet

Per SegmentBoundary Conditions

Isentropic Efficiency

Total Temperature Ratio

Total Pressure Ratio

Output Parameter



Mass Flow Rate

Rotational Velocity



Total Temperature Ratio


Mass Flow Rate





Optimization

Isotropic Efficiency [%]

Rotational Velocity [U/min]

Mass Flow Rate [g/s]

88.1455060.4EA

88.0475060.7ARSM

87.9480061.2Best of DoE

87.5500060.0Initial Design

Input Parameter:

Mass Flow RateRotational Velocity

Target:Isentropic Efficiency = MAX


Pressure Distribution

Initial Design Design of Experiments

Adaptive Response Surface Evolutionary Algorithm


Mach Number, Blade to Blade Plot

Initial Design Design of Experiments

Adaptive Response Surface Evolutionary Algorithm


Summary

• Improvement of Isentropic Efficiency +0.5%

• Axial Turbine has a global maximum of Efficiency

• Strategy 1 (DoE+ARSM):

– Number of Designs 53 (DoE=22, ARSM=31)

– Efficient to find “the maximum”

• Strategy 2 (EA):

– Number of Designs 105

– Finds “the maximum” in parameter space


Thank You!

design optimization of flow path with ansys …© 2008 ansys, inc. all rights reserved. 1 ansys,...

Documents