AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES ATUL KOHLI, NOVEMBER 8TH 2018 PREPARED FOR THE 17TH JET ENGINES SYMPOSIUM, TECHNION UNIVERSITY

• A UNITED TECHNOLOGIES COMPANY

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

2

INTRODUCTION

Dr. Atul Kohli

Senior Technical Fellow

Heat Transfer – Analytical Methods

Pratt & Whitney

Fellow of the ASME

21 years in industry, all at Pratt & Whitney

More than 30 refereed publications, 15 issued patents with over 30 pending

Expertise in heat transfer and cooling for hot section components

Drive improvements in design system and analysis capabilities for prediction of metal temperatures

BSME from the Indian Institute of Technology

MSME and PhD in Mechanical Engineering from the University of Texas at Austin

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

3

SIMULATION HAS MADE SIGNIFICANT IMPACT ON ENGINE DESIGN

1960 1970 1980 1990 2000 2010 2020 3D Euler + Boundary Layer

3D RANS

Multi-stage RANS

Unsteady Multi-stage Euler, RANS

Unsteady Multi-stage RANS

HLES & Multi-discipline

JT8D PW4000 GTF

F100 F119 F135

Wasp

Image credits unless mentioned: Pratt & Whitney

“75 percent of the manufacturing cost is committed by

the end of conceptual phase of the design process”1

1 D. Ullman, The Mechanical Design Process, McGraw-Hill, New York, 1992

0

10

20

30

40

50

60

70

80

90

100

Time

% c

om

mit

ted

Preliminary &

Detailed Design

Conceptual

Design

Mfg Cost

Resources

Decisions made here,

with relatively few

resources, are critical

Pull fidelity forward in

design cycle

HUGE OPPORTUNITY FOR SIMULATION TOOLS

10

20

30

40

50

60

70

80

90

100

70%

5%20% 5%

Percent

Influence

On Total

Cost

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

4

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

5

ENGINEERING SIMULATIONS ARE INTEGRAL PART OF ENGINE DESIGN

• CFD relatively mature for single component,

on-design analysis.

• Going forward, we need to model component

interactions and off-design flow physics.

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

6

SUCCESSFUL COMPONENT SIMULATIONS DRIVE METRICS AND COST

Legacy V2500 PW6000 NGPF

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

7

DESIGN PROCESS HAS EVOLVED WITH COMPUTING CAPABILITY

High fidelity large

assemblies

Increasing

Multi-discipline

Automation

Aero

Variation Structures

Small-to-Mid Scale Design for Variation (DFV)

DFV on High Fidelity Models

Faster, Higher Fidelity Analysis

High Fidelity

Optimization

Automation

Statistical

Methods

Component

models

Geometry

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

8

MULTI DISCIPLINARY DESIGN OPTIMIZATION IS KEY

Parametric Geometry

Automated Meshing Aerodynamic

CFD

Internal Flow CFD Film Cooling

Flow CFD

Internal Flow

Film Cooling

Internal H’s

Temperature Solution

Stress Solution

Str

ain

T

Life Prediction

MDO-Ready

Design System

MODELING

• Accuracy

• Parametric models

• Fidelity level vs run-time

AUTOMATED WORKFLOW

• Flexible

• Robust

• Easy to use

• Numerous variables/outputs

INFRASTRUCTURE

• Distributed, parallel computing

• Network reliability

• Data storage and search

CULTURE

• Alignment

• Up-front investment

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

9

MDO ENABLES GOOD DECISION MAKING EARLIER IN DESIGN CYCLE

Y

X

Maximum Fuel Efficiency

Minimum Cost

Design Space Map

Flowpath Selection

DOE

Parametric

Flowpath

Meanline

Analysis

Rotor Sizing

Cooling Air /

Life

Cost

Weight

System

Metrics

1000s of

analyses/day

Airfoil Optimization

DOE

Parametric

Airfoil

CFD (Aero-Thermal)

FEA (Structures)

Lifing

Part-Level

Metrics

10s to 100s of

analyses/day

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

10

PART LEVEL CHALLENGE- PREDICTION OF FILM COOLING FLOWS

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20 25 30

x/D

h

Present Study

Kohli and Bogard (1995)

M=0.5, DR=1.6, L/D=4

Pedersen, Eckert, and Goldstein

(1977) M=0.52, DR=1.5, L/D=40

Schmidt, Sen, and Bogard (1994)

M=0.6, DR=1.6, L/D=4

Sinha, Bogard, and Crawford (1991)

M=0.5, DR=1.6, L/D=1.75

Walters and Leylek (1996)

M=0.5, DR=1.6, L/D=4

Kohli and Thole (1997)

Notional RANS

Prediction

Notional

Data

• RANS methods not capable of predicting film cooling effectiveness due to turbulence model limitations

• LES shows promise but not practical for design cycle

RANS vs. data

© ASME, from ASME proceedings of 32nd National Heat

Transfer Conference, Vol 12, pp. 223-232

© 2009 by CERFACS, from VKI

Lecture Series on LES

Applications, page 25.

LES example

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

11

SOME SUCCESS USING OPTIMIZATION - TRENDWISE ACCURACY

Baseline

Optimized

-4

-2

0

2

4

6

8

10

12

14

-2 3 8 13 18 23

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1st Suction Row with Optimized GeometryDR = 1.6; M = 1.03

-4

-2

0

2

4

6

8

10

12

14

-2 3 8 13 18 23

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1st Suction Row with Baseline GeometryDR = 1.6; M = 0.94

-4

-2

0

2

4

6

8

10

12

14

-2 3 8 13 18 23

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1st Suction Row with Optimized GeometryDR = 1.6; M = 1.03

-4

-2

0

2

4

6

8

10

12

14

-2 3 8 13 18 23

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1st Suction Row with Baseline GeometryDR = 1.6; M = 0.94

• An optimization procedure using

RANS used to change airfoil

shape with film cooling.

• Increase in adiabatic effectiveness

with no impact on aerodynamic

losses.

• Increased local acceleration

and convex curvature reduced

blow-off and improved lateral

spreading.

• Test data showed similar

trends as predictions but not

absolute magnitudes.

• Despite its limitations, RANS

methodology can be useful!

© ASME, from Kohli & Bogard, GT2006-90852

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES 12

COMPONENT LEVEL CHALLENGE – PREDICTING INGESTION

• Understanding gas path – rim cavity interactions critical for both compressors & turbines

• Multi-row, ‘full-wheel’ time accurate simulations required to capture complex ingestion phenomena

Complicated pressure field changes

with time/position

Complex interaction between

gaspath/ingested/purge flow

Full-wheel domain for

analysis

© ASME, from Wang et al., GT2012-68193

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

13

MULTI COMPONENT CHALLENGE – COMBUSTOR TURBINE INTERACTION

• Different modeling fidelity used in individual components (LES in combustor vs. URANS in turbine)

• Large differences in length/time scales of interest (cooling air vs. gas path) makes full turbine LES simulation prohibitive

Stanford University, 2006 LES RANS

• More reliance on simulations, from cradle to grave – digital thread/twin

• Need for both high and low fidelity modeling – machine learning

• Continued focus on multi-disciplinary and design for variation

AERO-THERMAL METHODS FOR ENGINE DESIGN: EVOLUTION AND CHALLENGES

14

WHAT’S ON THE HORIZON?