industrial applications of multidisciplinary design … applications of multidisciplinary design...

34
Industrial Applications of Multidisciplinary Design Optimization Karthik Chittepu CADFEM 2 nd Optimization and Stochastic Days December 3-4, 2012

Upload: vantu

Post on 09-Mar-2018

218 views

Category:

Documents


1 download

TRANSCRIPT

Industrial Applications of Multidisciplinary Design Optimization

Karthik Chittepu

CADFEM

2nd Optimization and Stochastic Days December 3-4, 2012

-1-

Industrial Applications

Consumer Goods

Automotive

Health

Aerospace & Ship Building

Construction & Materials

Energy

-2-

Technical Challenges

Design to the point New Power train technology Comfort – (N.V.H, etc)

Safety – (crash, etc)

Weight

Paint Process

New Materials

composites

High strength alloys

polymers

Emissions

Automotive Applications

-3-

Automobile Applications: Search for Alternate Vehicle Concept

• Search for alternative full vehicle concepts by means of local and global optimization strategies.

• The car body are formulated by numerous performance constraints - crash, stiffness and driving comfort requirements.

• Only extremely small areas of car dimensioning exist which are admissible and interesting.

• The optimization task does not only consist of an optimization of known car concepts, but search for alternative vehicle concepts.

• Using optiSLang for the search for alternative concept and weight optimization on the islands of new concept help fulfilling all constraints - significantly lower weight.

• 6% (65 kgs) reduction in weight was achieved with optimized new vehicle concept.

Fig: Full Car Model at Concept Phase

Fig: History of One of the Constraints

Fig: Weight Difference between Models Courtesy: BMW

-4-

Automobile Applications: Parametric Optimization of Oil Pan

• The aim of this project was to improve the acoustic behaviour of the oil-pan by keeping the mass on a relatively constant level.

• For development, topology optimization or bead optimization is used frequently for mass-reduction and stiffness increment. The user does not define the parameters explicitly in these methods.

• Due to this, problems occur in transformation of the design to manufacturing and also to robustness check.

• Sensitivity analysis was carried out to reduced the bead parameters from 32 to 13 for optimization.

• 50% increase in Eigen frequency was achieved and design was implemented in manufacturing process.

• Robustness check was carried to check the performance of the product.

Fig: Initial Oil Pan – without Beads

Fig: Optimized Oil Pan

Fig: Realized Oil Pan for Manufacturing

LL125, BL = 1mm

0

5

10

15

20

25

30

-0.80 -0.60 -0.40 -0.20 0.00 0.20 0.40 0.60 0.80

Messstrecke [m]

Sch

icht

dick

e [µ

m]

MessSimulation

-5-

Automobile Applications: Optimization of Process Parameters for Paint Application

• Quality of paint is important – Corrosive resistance and quality impression (marketing).

• BMW has developed State of Art procedure to optimization the paint parameters using ANSYS Fluent and optiSLang thus reducing the development time for new car by 50%.

• This approach predicts the paint thickness with high accuracy compared to measurements .

• 5 Input parameters are considered based on the paint process

Painting distance

Paint mass flow

Rotational velocity of the paint bell

Strength of the electrical field

Mass flow rate of the guiding air

Fig: Paint Process During Manufacturing Phase

Fig: Comparison of Test and Simulation

Fig: Paint Thickness Distribution as Contour

Courtesy: BMW

50%

m_max

sp50 mu_diff

m_min

-6-

Automobile Applications: Optimization of Process Parameters for Paint Application

• In contrast to the result of a real configuration, the ideal paint distribution isshaped like a cylinder with a fixed radius.

• Sensitivity analysis carried out to enhancement the understanding of the effect of the parameters on the paint process.

• Optimization has been carried out to achieve significant improvement in the pain process.

Fig: Thickness distribution of all designs

Fig: Thickness Distribution of Optimized Design

Fig: Thickness Distribution of Initial and Ideal Case Courtesy: BMW

-7-

Automobile Applications: Global Sensitivity Analysis of GDI Nozzle

• According to the European policy for CO2

regarding new car registrations, the CO2 targets have to be reached step by step on average from 2012 until 2015 as shown in the graph.

• Soot emission and CO2 emission are affected by the injection pressure.

• Dimension of the nozzle has effect on the injection pressure. Therefore they are defined as the input parameters.

• From sensitivity analysis important input parameters effecting the particular output can be determined.

• Also worked out the improvement potential and direction.

• Predicted trends have been confirmed by spray and engine experiment results and methodology was implement into nozzle design development cycle.

Fig: European CO2 Emission Targets

Fig: Nozzle Parameter and CFD model

Fig: CoP and Meta Model of Injection Velocity Courtesy: Continental

Inner hat No.3

Inner hat No.2 Inner hat

No.1

-8-

Hat

Plate

Inner hat

Added mass

64[km/h]

Rigid Wall

Rigid Plate

Objective value & Constraints

Objective value

• Displacement of rigid plate

Constraints

• Acceleration of rigid plate :

Acceleration(new)<Acceleration(initial)*0.95

• Mass of inner hat : Mass(new)<Mass(initial)

Automobile Applications: Optimization of Crash Beam

Courtesy: TECOSIM Japan

-9-

ID61 (Best Design) Initial Design

0 [msec]

70 [msec]

Automobile Applications: Optimization of Crash Beam

Courtesy: TECOSIM Japan

-10-

Automobile Applications: Optimization of Crash Beam

Mass

Dis

pla

cem

ent

Best Design ID 61 Best Design

Fail to satisfy constraint

Initial Design

ID 5 ID 11

ID61(Best Design ) 380.8[mm]

Initial Design 497.8[mm]

ID 11 213[mm]

ID 5 211.3[mm]

ID61 (Best Design)

Initial Design

ID 11 ID 5

Fig: Energy Curve

Fig: Displacement curve

Courtesy: TECOSIM Japan

-11-

Automobile Applications: Modern Parametric Process for Brake-Squeal Simulation

• Brake is an important and complex, safety and performance component in automobile.

• NVH field complaints and warranty costs as well as permanently increase in the customer requirements and targets

• Required features for NVH simulation process

Efficiency

Result Quality

Sensitivity / Optimization / Robustness process

-12-

Automobile Applications: Modern Parametric Process for Brake-Squeal Simulation

• Geometry (rib height and position) and material data (Young’s Modulus in thickness direction, frictional values) are parametrized for investigation

• This kind of closed loop technique will lead to a efficient and accurate results. Therefore helps in establishing a Simulation Driven Product Development .

-13-

Automobile Applications: Optimization of Main Relay for Hybrid Car

• LEV200 main relay for hybrid truck application.

• Relay damaged due to levitation - Increased pressure inside relay cause of explosion.

• Levitation Effect on Contact Bridge - Spring force reduced by force caused by short circuit.

• Arc will blow out to outer side.

• Objective was to minimize both the force and material in magnet which are conflicting.

• Due to this Pareto optimization is used.

• Two best design are show in the table. Based on the requirement of the designer, design is selected for manufacturing.

• This process was integrated into the design cycle.

Fig: Main Relay of Hybrid Car

Fig: Mathematical Model of Main Relay

Fig: Best Designs from Pareto Optimization Courtesy: TE Connectivity

Design 2 Design 8

F arc 0.854 (0%) 0.877 (+3%)

F Bridge 11.1 (-3%) 11.33 (-1%)

Magnetic volume

2968.7 (-29%) 1974.1 (-53%)

-14-

Construction & Material Applications: System Identification of High-Speed Railway Bridges

• In the case of an enhancement of an existing line, it is required to proof all bridges with respect to resonance problems and is usually performed by numerical analyses.

• The correct modeling of an existing bridge is a complicated task.

• Numerical model are updated based on the experimental obtained results to represent realistic behavior.

• Sensitivity analysis is performed to reduce the input parameter set and then optimization is performed to fit the responses

Fig: Isometric View of the Finite Element Model

Fig: Comparison of the Natural Frequencies and Mode Shapes Courtesy: Bauhaus-University Weimar

Fig: Results of the Optimization

-15-

Construction & Material Applications: Parametric Identification of Gurson Material Model

• The requirements on passive safety systems, fuel efficiency and co2 reduction have grownto high standards - increase in the use of high strength steels and simulation accuracy.

• For crashworthiness simulation - description of the dynamic material behaviour combined with failure prediction -> Gurson Model.

• Gurson model is based on a micromechanical model describing growth and nucleation of spheroid voids.

• Damage parameter (EN, FC, FF0, F0, SN, FN) effect the Gurson material model – highly expensive and complex to identify from tests.

• Parameter identification process was adapted to update Gurson parameters - fitting the simulation data to experimental/test data.

Fig: Tensile Test

Fig: Comparison of Stress Strain curve

Fig: Parametric Data of Best Design Courtesy: TATA Motors

-16-

Construction & Material Applications: Optimization of Visco-Elastic Models of Mold Compounds

• Microelectronic packages exposed to temperature load.

•Multi-Layer structures composed of complex materials.

• This analysis is set up in ANSYS which require more than 20 parameters for the Visco - Elastic model – difficult to obtain data from tests

• Tensile test is carried out under sine load. The tests are carried out at different temperatures and frequencies.

Fig: Tensile Test

Fig: Comparison of Stress Strain curve

Fig: Parametric Data of Best Design

Courtesy: Qimonda

-17-

Construction & Material Applications: Optimization of Visco-Elastic Models of Mold Compounds

• 41 Input parameters are defined – E0, E1, n, C1, C2 and ai (2*18)

• From sensitivity analysis it is observed that E0, E1, C1 are high influence and C2 and ai

G also have influence and rest have no influence on the response parameter – not selected for optimization

• Optimization was carried out significant improvement was achieve in comparison to trail and error method

Fig: Comparison of Stress Strain curve

Fig: Best Design based on the optimization using optiSLang

Courtesy: Qimonda

Fig: Initial design using conventional approach (trial and error combined w/ experience).

-18-

ANSYS Mechanical

ACP Post processing

Model preparation in Design modeler

ACP Preprocessing

ANSYS (Batch-Run)

New Parameter set in optiSLang

Mesh

Layer orientation

Post processing in optiSLang

Position of Layers

Layer thickness

Boundary Conditions

IRF

Deformation

optiSLang routine

Material Applications: Laminate Composite Material Optimization

-19-

Aerospace and Ship Building Applications:

Courtesy: Airbus

-20-

Aerospace and Ship Building Applications: Optimization of Higher Crash Load of Sail Plane

-21-

Aerospace and Ship Building Applications: Optimization of Higher Crash Load of Sail Plane

Fig: Parametric Data of Best Design

• Geometric modeling using ANSYS-Workbench

• carbon fiber material definition (Stiffness/strength/damage) using ANSYS Composite Modeller (ACP)

objective: minimum Mass

constraints: maximum load, no critical damage of structure (IRF -values)

With optimization of position, orientation and thickness of important fabric layers the Load could be improved by 50% having a mass increase 1.6% only!

-22-

Aerospace and Ship Building Applications: Robust Design Optimization on a Gas Turbine Component

• Optimization, robustness and reliability analyses have an increasing importance in aviation industry engineering.

• The turbine consist of several stationary nozzle guide vanes and row of rotating blades. Those blades are supported by a disc. A disc of the high pressure turbine (HPT) is the object used in this application.

• The analyzed disc is subjected to manifold thermal and mechanical loads. Analysis is carried out using ANSYS Workbench.

• An axis symmetric 2D-model was deemed to be a sufficient compromise for result accuracy and analysis time.

• Mass and Life cycle number are the two design objectives.

Fig: Gas turbine engine

Fig: Axis symmetric model with schematic loads and boundary conditions

Courtesy: Rolls-Royce

-23-

Aerospace and Ship Building Applications: Robust Design Optimization on a Gas Turbine Component

• Sensitivity analysis has carried to estimate the region where occurrence of minimum life cycles have high probability.

• The investigation deals with two conflicting objectives: predicted life and mass of the disc. Hence a multi-objective optimization had to be performed.

• This design decision normally is made based on higher-level information, e.g. customer requirements for a certain flight mission.

• A design with a nominal expected life of 35280 cycles was chosen, which lead to a mass of about 35kg.

• Reliability analysis was carried out to check the probability of failure due to naturally occurring scatter.

Fig: Partitioning of disc to identify location of minimum life.

Fig: Results of reliability analysis

Courtesy: Rolls-Royce

Fig: Pareto Frontier

-24-

• Optimization of the total weight of two load cases with constrains(stresses)

• 30,000 discrete variables

• Self regulating evolutionary strategy

• Population of 4, uniform crossover for reproduction

• Active search for dominant genes with different mutation rates

Solver: ANSYS

Design Evaluations: 3000

Design Improvement: > 10 %

Aerospace and Ship Building Applications: Robust Design Optimization on a Gas Turbine Component

-25-

Energy Applications: Optimization of gas production

• Natural gas and oil reservoirs are often located in layered rock formations with low permeability like the Barnett Shale in Texas, US.

• Hydraulic fracturing is used routinely in gas and oil industry to create a large network of permeable fractures.

• It creates a large network of permeable fractures which connects the production well with the greatest possible volume of reservoir rock for profitable gas production rates

Fig: right: image log of well will fracture characteristics

left: core picture with bedding plane and joint

Courtesy: Rolls-Royce

Fig: Image log measurement data of the 7 rock layers

-26-

Energy Applications: Optimization of gas production

• For an effective three dimensional analysis was carried out using ANSYS® and multiPlaswhich simulates the three dimensional fracturing of rock and reopening of joints during the fracture growth.

• The simulator was set up and verified for two wells in the Barnett Shale production area. Available first core measurements are used to calibrate the reservoir model and tested with second core measurements.

• Goal of the optimization of the hydraulic fracturing procedure is maximizing the fractured reservoir rock volume which results in the maximization of gas production. 25% increase in production was achieved

Fig: Seismic fracture mapping measurement

Fig: Stimulated rock body after 193 minutes of pressuring

-27-

Consumer Good Applications Optimization of a Date Mechanism

• Watch industry mechanisms involve a large number of high precision flexible pre-constrained mechanical components.

• Using traditional prototyping, the definition of non-deformed geometries for production is a costly manual iterative process.

Fig: Gas turbine engine

Courtesy: AUDEMARS PIGUET

• Optimization of the mechanism that allows changing the date display every 24 hours with 0.015s and only change of one day.

-28-

Consumer Good Applications Parametric Optimization of Washer Bellows

• Bellow connects moving washing unit (tub + drum) and stable door opening (insert) and as sealing against water and steam inside drum during wash cycle

• Main objective is to keeping transmitted forces as low as possible, decreasing of noise, guaranteed sealing properties for whole lifetime of washer (durability)

• Transmitted forces from washing unit to cabinet washer impact stability of washer

• Wearing out during lifetime is due to abrasion caused by self-contact inside bellow and consequently to holes through which water and steam can leak out

Courtesy: Whirlpool

-29-

• Horizontal force decreased by ca. 31%

• Vertical force decreased by ca. 22%

• Eliminated contact forces

•Such bellow would transmit lower forces to cabinet (lower vibration and noise) and it’s durability is much higher.

Courtesy: Whirlpool

Consumer Good Applications Parametric Optimization of Washer Bellows

-30-

Health Applications: Optimization for Development of Vascular Stents

• Atherosclerosis is plaque build-up inside the coronary arteries which is the most commontype of heart disease.

• One of the most important achievements of the last years in interventional cardiology.

• Closely integrated CAD + FEA is used from the early development phase on.

• Simulation is carried out for 18 load cases.

Courtesy: CORTRONIK

• Geometry is parameterized and optimization is carried out

• “1 year intense usage & experience of optiSLang proofed to be very successful and highly effective” CORTRONIK Stent Development

-31-

• Solver: ANSYS • (using automatic spot weld • Meshing procedure) • Design evaluations: 200 • Design improvement: 47%

2

)( /140cossinsincos mmNMYMXFZFYFXR

• 134 binary variables, torsion loading, stress constrains

• Weak elitism to reach fast design improvement

• Fatigue related stress evaluation in all spot welds

Other Applications: Optimization of Spot Welds

-32-

Other Applications: Optimization of Sea Hammer

Dynamic performance optimization under weight and stress constraints using 30 CAD-parameter. With the help of sensitivity study and optimization (ARSM), the performance of a deep sea hammer for different pile diameters was optimized.

Initial Design valid for two pile diameter

Optimized design valid for four pile diameter

weight=4365 kg weight=5155 kg

Design Evaluations: 200 times 4 loadcase CAE: ANSYS workbench CAD: ProEngineer

Thank You