Morphing process for CFD without restrictions

Headquarter Research and TechnologyZ-FB / Central EngineeringMorphing process for CFD without restrictionsJ. Jasper, Dr. P. Seggewiß, KSPG AG, R. Schwarz, Dr. A. Veitl, ALTAIR Engineering GmbH

EATC 2013 Turin, Italy, April 24th 2013

www.kspg.com © KSPG AG 2013

Morphing Process for CFD without restrictions

Authors:

Mrs. Jessica Jasper Mrs. Ramona Schwarz [email protected] [email protected]

Mr. Dr.-Ing. Peter Seggewiß Mr. Dr.-Ing. Armin [email protected] [email protected]

KSPG AG ALTAIR Engineering GmbHAlfred-Pierburg-Str. 1 Calwer Str. 741460 Neuss 71034 BöblingenGermany Germany

Abstract

The identification of an appropriate concept for a required system performance is a typical task for technical design and CAEbased simulation. The typical work share is that CAD defines the geometry and CAE is calculating the performance. A more modern alternative is a process which is fully embedded in CAE. Those processes are usually faster than partly or non-embedded processes and they require less coordination effort which helps to improve development performance. Morphing for structural mechanics is a well-established method to support such modern CAE-processes. Unfortunately, used in CFD environment they result in significantly reduced handling performance. The reason is the higher numbers of elements. Another drawback of the usage of morphing in CFD environment is the highly limited freedom for robust geometrical changes.

To overcome these issues, KSPG AG and Altair Engineering developed a new morphing process which leads to nearly nodesign restrictions and respects the expectations for an interactive and creative work flow including assessment of theconcept ideas. This gives the high flexibility to integrate the new process in an automated system optimization, which wassuccessfully proven for an exhaust gas recirculation example.

Headquarter Research and Technology –Morphing process for CFD without restrictions


2


Agenda

Introduction1

New morphing process for CFD2

Summary3

Outlook4



3


Agenda

Introduction1


Summary3

Outlook4



4

Den Präsentationstitel hier übernehmen und in der Masteransicht eingeben

www.kspg.com © KSPG AG 2013 5www.kspg.com © KSPG AG 2013 5

A part of the Rheinmetall Group

European Defence company forground forces technology

Automotive supplier of enginecomponents and systems

Sales: €4.704 billionEmployees: 21,767



Rheinmetall DefenceKSPG AG

Data based on fiscal 2012

Rheinmetall AG

KSPG part of the Rheinmetall Group

International Partner forSecurity and Mobility



Three strong divisions

Hardparts Mechatronics Motorservice

Div

isio

nsB

usin

ess

Uni

ts

Pistons

Aluminum-Technology

Plain Bearings

Large-Bore Pistons

International

Domestic

Pierburg

Pierburg Pump Technology

The structure of the KSPG Group



Division Hardparts

Pistons modulesfor passenger cars

Power cylinderunits for trucks

Continouscasting

Dry bearings(Permaglide®)

Steel pistons forpassenger cars

Aluminum engine blocks Transmissionhousings

Cylinder workingsurface coating

Cylinder heads

Crankshaftbearings

Bearings forcommercial vehicles

Passenger carspistons

Plain bearings,bushings

Large-bore pistons in the diameter range from 150 to 640 mm made of steel and aluminum

Pist

ons

Alu

min

um-

Tech

nolo

gyPl

ain

Bea

rings

Larg

e-B

ore

Pist

ons

Hardparts

Steel pistonsfor truck



Electrical trans-mission oil pumps

Division Mechatronics

Secondary airpumps

Variable oil pumps

Compact-EGR-valves

Elect. motor drivenexhaust flaps

Electrical coolantpumps CWA 400

Fully variable waterpumps for trucks

EGR- coolermodules

Single-vanevacuum pumps

EGR valves forcommercial trucks

Exhaust gasmass flow sensors

Pier

burg

Pier

burg

Pum

p Te

chno

logy

Water circulationpumps

Mechanical oilpumps for trucks

Turbo boost re-circulation valves

UniValve: Mechan.Valve control system

Mechatronics

Electrical coolantpumps CWA 50

Elect. throttlebodies for trucks

EGR mixermodules

Mechanical waterpump with clutch

Mechanicalwater pumps



Motorservice

Sales organisation for the worldwide aftermarket activities of the KSPG Group

Independent Aftermarket

OES (OE Service) for KS and PIERBURG brands

SOE (Special OE) for KS and PIERBURG brands

Division Motorservice

International

MS Motor Service Deutschland

MS Motor Service France

MS Motor Service Asia Pacific

MS Motor Service Iberica

MS Motor Service Brazil

MS Motor Service Istanbul

Domestic

BF GermanyAsperg

MS Motor Service International


Optimization Tools Introduction: Goals

Enabling of simulation driven design optimization according to requirements.

Reduction of design loops in structural and fluid design by fast and precise results.

Reach optimum usage of design space and full creative leeway.

Support reduction of development times:

Realizing frontloading in product development. Facilitating feasibility studies, risk analysis and

cost appraisals. Reduction of development loops. Reduction of prototype costs.

Support reduction of development costs:

Better design to customer requirements. Better structural performance with equal or even

less product weight. Better flow characteristics (less pressure losses,

better uniformity, ...).

Support improvement of product quality:

Market needs:

Timing

Costs

Quality

Future

Initial situation

Today

10




Optimization Tools Introduction: Status

For Mechatronics introduced since 2007 and enhanced in 2008 and 2009. All structural CAE engineers for Mechatronics products were trained. For Hardparts introduced since 2011 and enhanced in 2012. Key structural CAE engineers for Hardparts were trained. Regular usage if suggested or by request.

Structural Optimization Tools:

For Mechatronics introduced since 2007 and enhanced in 2008 and 2009. All CFD engineers for Mechatronics products were trained. Introduction for Hardparts started in 2012. But, very poor usage due to specific product requirements and CFD related

obstacles.

Fluid Optimization Tools:

Formal integration in development processes have been easy, real usage of the tools still remains difficult, therefore additional promotion and management support has been needed.

And additional effort for CFD optimization had to be spent to obtain the full range of results:

The implementation process needs to be accomplished, to get fast and precise proposals opening new design options and full creative leeway:



11


Agenda

Introduction1

New morphing process for CFDProject background2

Summary3

Outlook4



12


Project Background I

Classical work sharing between CAD and CAE: CAE engineer provides results including proposals for geometry improvements, designer modifies the CAD data.

Advantage: Fast and significant design changes are possible.

Disadvantage: Challenging interface management between CAD and CAE. High efforts for coordination and communication in order to keep error-proneness low. Creative and interactive optimization in CAE is not possible.

New approach: Ideas are generated directly in CAE. A creative and interactive optimization process is possible.

Where do we come from?

CAE Modelling

CAD Design Simulation ManufacturingConcept

IdeaSystem

Evaluation TestingCAD Design

CAE Modelling SimulationConcept

IdeaSystem

EvaluationCAD

Design ManufacturingTesting



13


Project Background IIa

Classical HyperMorph approach: morphing of CFD volume mesh. Challenge: control deformation of prism layers during morphing!

Advantage: No change of software tool environment, executable as batch process.

Disadvantage: Due to very high element counts (which are usual for KSPG AG CFD tasks) significant performance

reduction compared to structural mechanics application. High efforts for prism layer control.

MORPHING of CFD grid using HyperMesh and HyperStudy

1140 morph volumes1296 handles

2D-domains for prism layer control

14




Project Background IIb

Major issue of CFD when applying classical HyperMorph approach: Highly limited freedom for design changes → significant modifications cause mesh damage! Scatter band of numerical uncertainties in CFD is equal or greater than the influence of (limited) geometry

modification.

Application of classical CFD mesh morphing is not usable/makes no sense!

Previous attempts to solve the problem were not executable in HyperStudy batch process.

Release of KSPG AG & ALTAIR Project “New Morphing Process for CFD”.

Red elements indicate mesh damage!Initial Situation before morphing. After applying classical morph approach:

Schematic example for mesh damage:

15




Agenda

Introduction1

New morphing process for CFDCharacteristics of new morphing process2

Summary3

Outlook4



16


New Process for CFD Morphing – What is different?

Shapes are applied on Volume Mesh. HyperStudy combines 2 ASCII files: (1) displacement vectors (2) grid point coordinates.

Major development target: The approach for "New Process for CFD Morphing" is based on batch morphing and batch

meshing (2D & 3D). Challenge: ensure a robust and reliable process considering all CFD related requirements.

Ideas for target realization: Splitting the complete system into short and handy sections (domains). Morphing and batch meshing will be applied only on one single section (→ morph domain). New morph process will be supported by scripts. Automated model export and assembly.

Initial situation “Classical Morphing”:

17




New Process for CFD Morphing – Main Procedure

Small and handy submodels for quick and interactive modifications.Partitioning & meshing: Divide system in smaller sections (domains). Precise interface definition. Individual 2D and 3D meshing. Export of each section separately.

Morphing: Apply standard morph functions (shape functions / shape variables) at outer 2D mesh only. Apply new automated CFD morph process (in batch mode) with:

Remeshing of morphed outer surface (2D). CFD-Remeshing of inner volume (3D).

Domain assembly & job submission: Automatic assembly. Automatic load case setup.

Basic idea:

2D

3D

18




New Morphing Process in an Optimization Environment

Input definition and preparative work. HyperStudy Setup. Automated CFD morph process in batch mode inside HyperStudy.

STEP 2 STEP 3

Workflow:

nominal run automatic morphing,remeshing, domain export

assemblyjob

submissionresult

processing

control/optimizationengine

HyperStudy

all domains meshed morph shapes defined submission scripts prepared

STEP

1

19




Agenda

Introduction1

New morphing process for CFDExample: flow through reed valve housing 2

Summary3

Outlook4



20


EGR reed valve: Assembled downstream of EGR cooler. Makes use of the wave action in the exhaust and

intake system to push the EGR rate. Aluminum and PPA reed valve available.

CFD task: Reduce swirling flow in reed valve housing: Pressure loss reduction. Increase flow uniformity downstream of reed valve

housing.

DoE Setup: 1 load case (testing conditions). 5 shapes which modify the contour of reed valve

housing.Load Case Definition

Mass Flow 70 g/s

Inlet Temperature 20 °C

Outlet Pressure 1 bar (stat., abs.)

Fluid Material Air Ideal Gas

Alu

min

um

PPA

Example: Flow through Reed Valve Housing

21



www.kspg.com © KSPG AG 2013 22

One big model.Difficult to handle, to mesh and to modify in CAE.

Domain partitioning: Each domain is easy to handle, to mesh and to modify!

Standard meshing without domain partitioning:

> 15 Mio. Elements

assembly tool

4 x (< 5 Mio. Elements)

Assembly on Demand = Model on Demand


STEP

1



STEP 1: Domain Partitioning & Meshing


STEP 1: Morphing & CFD Batch Meshing

Creative definition of promising geometric changes based on engineering judgement.

Transfer of idea into CAE.

Definition of process settings in HyperMesh: Interface sections for later assembly. Wall components for boundary layer. Volume for 3D meshing.

Use GUI !

Morph Shape generation:


STEP

1

Mesh of Morph Domain Reduce ModelDefine

Design Space Fix Interfaces Morph 2D-Mesh Export Shapes

23




STEP 1: Batch & Submission Scripts

Define design space based on component names.

Define CFD meshing setup: Number of layers. First layer thickness. Growth rate.

Define 2D-remeshing of the design space: Type of elements.Quality of 2D-mesh is controlled by a standard criteria file which could be easily created with HyperMesh.

Possibility to switch between local run on workstation or submission of the CFD job to cluster by the use of PBS.

Script configuration:


STEP

1

24




5 Shapes for reed valve housing contour: Smooth radius at pipe inlet. Remove bump. Short side: push wall inside. Long side A: pull wall outside. Long side B: push wall inside.

STEP 2: Batch Morphing & Resulting Shapes

remove bump

Shape „Long Side A“: approx. 6mm

≈ 12

0 m

m

Morphing result: Deformations > 1mm are possible. Successful shape combinations. Not fully smoothed shapes:

Potential for Improvements!

short side

long side

A

B

25




Goals of postprocessing: Fully automated postprocessing. Only flow visualisation is not sufficient. Required: characteristics which represent flow

field structure. Provide data base for HyperStudy responses. Identification of best design in DoE.

CFD Postprocessing & Definition of HyperStudy Responses (a)

STEP 2: TEMPLEX

Streamlinescoloured by Velocity

Angle Deviation

Normal Velocity

26




Usage of TEMPLEX: Raw data (velocity components at grid nodes)

is taken into account. Statistical evaluation of velocity distribution →

HyperStudy responses. Almost independent of postprocessing

software. Full flexibility and short response time.

CFD Postprocessing & Definition of HyperStudy Responses (b)

STEP 1: export raw data for each node on planeX [m], Y [m], Z [m], u [m/s], v [m/s], w [m/s],

STEP 2: TEMPLEX

RESULT: output.txtCharacteristics A1, B1, C1 etc.

HyperStudyResponses

27




35 CFD results from fullfactorial DoE: 25 (5 Shapes, 2 Levels) → 32 Runs. 3 additional Runs (3rd Level for 3 Shapes). HyperStudy Response: Characteristic for Swirl in Pipe.

Flow through Reed Valve Housing – DoE Results

Run

1

Angle Deviation

Less Swirl in Pipe! Run

33

High

Low

28




Agenda

Introduction1


Summary3

Outlook4



29


Summary

Classical mesh morphing approach in HyperMesh was not applicable for CFD optimization tasks with regard to KSPG AG products.

Source of problem: Highly limited freedom for design changes → significant modifications cause mesh damage. Scatter band of numerical uncertainties in CFD ≥ influence of geometry change.

Approach: Release of development project together with ALTAIR Engineering GmbH. Definition of a new robust process for CFD morphing:

Domain partitioning. Batch process for morphing of surface mesh & CFD meshing. New postprocessing procedure.

Fast and reliable predictions for significant geometry modifications. Comfortable handling of workflow by HyperStudy user. New process successfully applied for reed valve application.

Numerical Optimization for CFD at KSPG AG has been fundamentally strengthened !

Problem:

30




Agenda

Introduction1


Summary3

Outlook4



31


Outlook

Application of new CFD morphing process in daily project life.

Embedding of further functionalities: Exchange of (non-morph) domains in order to explore a wider range of geometry

variants during DoE. Modification of CFD boundary conditions (different load cases).

KSPG AG products require a process for morphing for CFD CHT tasks !!!

Modification of Coolant Channel

32



Thank you for your attention!


ENDE

Die Weitergabe sowie Vervielfältigung dieses Dokuments, Verwertung und Mitteilung seines Inhalts sind verboten, soweit nicht ausdrücklich gestattet. Zuwiderhandlungen verpflichten zu Schadenersatz. Alle Rechte für den Fall der Patent-, Gebrauchsmuster- oder Geschmacksmustereintragung sind vorbehalten.

The reproduction, distribution and utilization of this document as well as the communication of its contents to others without explicit authorization is prohibited. Offenders will be held liable for the payment of damages. All rights reserved in the event of the grant of a patent, utility model or design.

Copyright

34


Disclaimer

This presentation contains “forward-looking statements”. Forward-looking statements are sometimes, but not always, identified by their use of a date in the future or such words as “will”, “anticipates”, “aims”, “could”, “may”, “should”, “expects”, “believes”, “intends”, “plans” or “targets”. By their nature, forward-looking statements are inherently predictive, speculative and involve risk and uncertainty because they relate to events and depend on circumstances that will occur in the future. There are a number of factors that could cause actual results and developments to differ materially from those expressed or implied by these forward-looking statements. In particular, such factors may have a material adverse effect on the costs and revenue development of the KSPG group.

All written or oral forward-looking statements attributable to KSPG AG or any group company of KSPG AG or any persons acting on their behalf contained in or made in connection with this presentation are expressly qualified in their entirety by factors of the kind referred to above. No assurances can be given that the forward-looking statements in this presentation will be realized. Except as otherwise stated herein and as may be required to comply with applicable law and regulations, KSPG AG does not intend to update these forward-looking statements and does not undertake any obligation to do so.

35

Morphing process for CFD without restrictions

Technology

Transcript of Morphing process for CFD without restrictions