Improved Near Wall Treatment for CI Engine CFD Simulations

Mika Nuutinen Helsinki University of Technology,

Internal Combustion Engine Technology

2

Conjugate Heat Transfer in CFD

Continuous heat flux across surface Simultaneous determination of heat flow

and temperature within a fluid and its adjacent solid e.g.

• Cylinder charge and piston• Engine block and coolant

3

Why conjugate heat transfer?

Primarily: Designer needs accurate temperature data in/on solid part• Maximum temperature (melting)• Temperature distribution (thermal loads)

Secondarily: Produces transient, more accurate boundary condition for temperature• More accurate heat loss prediction• More accurate overall temperature/pressure fields

4

CFD problems in heat transfer

Inaccuracy of RANS turbulence models (k-ε, k-ω)

Extreme field gradients near walls

Standard wall treatment (wall functions) omits the effects of temperature induced density variations near walls

5

New wall function formalism Derived similarly to standard wall functions, but

with smooth turbulent viscosity transition (Mellor) and variable near wall turbulent Pr (Kays)

Sensitive to temperature induced density variation near the walls unlike standard wall functions+ Improves heat transfer and temperature predictions+ Easy to include other temperature variable effects to e.g. heat capacity, μ, k…- No analytical solution -> computational burden

6

Essential equations wt dy

du

wt

tp q

dy

dTc

PrPr

)_(/

/

21

41

,

,

termlamkCu

uyyy

ppwpf

pfppwp

pf

w

pw

w

u

uuuuu

,/

w

wpfppp

q

TTucT

,,

T

T

y

y

yT

T

yT

T

p

p

p

5.382

5.382

3

4

3

4

85.0Pr

7.0Pr t

1

dydu

Pr85.07.0

Pr

Pr

12

dydT

7

Velocity wall functions (hot gas case)

7.0Pr

1500

700

200

KT

KT

y

p

w

p

8

Temperature wall functions (hot gas)

7.0Pr

1500

700

200

KT

KT

y

p

w

p

9

Wall Heat flux prediction (hot gas)

7.0Pr

1500

700

200

KT

KT

y

p

w

p

10

Wall function comparison, typical CI engine simulation case

Simulations were made with 4 combinations of turbulence models and near wall treatments:

1) High Reynolds number k-e model with standard wall functions.

2) High Reynolds number k-e model with the new variable density wall functions

3) High Reynolds number RNG k-e model with standard wall functions.

4) Low Reynolds number k-e model with hybrid wall treatment.

11

Spray and Combustion modeling Lagrangian particle tracking Transfer of mass, momentum and heat

modeled Droplet break up models: Reitz-Diwakar

etc. Turbulent dispersion, collisions,

coalescence EBU LaTCT (laminar and turbulent

characteristic time) combustion model

12

Computational grid, fluid & solid zones

13

Cylinder pressure

14

Piston heat transfer

15

Piston peak temperature

16

Piston surface temperature

17

Concluding remarks The new wall function formalism works

well in practical simulations Enhances the predicted wall heat

transfer in CI engine simulations when the gas is hot (and vice versa)

Further improvements easy to implement

Computational burden can be minimized by selecting a smaller boundary where the heat transfer is critical