Researchers : Youngjae Kim, Eunbeom Jung, Taeksoo Kim, Sanghyuk Nam, Hakseon Lee, Seongwon Kang
Motivation
- Turbulent Prandtl number is an important physical parameter for predicting turbulent heat transfer.
- In RANS models, turbulent Prandtl number is often assumed as a constant based on Reynolds analogy and mixing length theory.
- However, turbulent Prandtl number shows large spatial variation for complex flows as observed in previous experimental
and numerical studies.
- It can cause inaccurate prediction of heat transfer rate in RANS simulation.
Objectives
- To analyze turbulent Prandtl number varying over space in various turbulent flows.
- To obtain database for data-based modeling using artificial neural network.
DNS for turbulent wedged channel
- Flow configuration & geometry specification
- Vortical structures & contours of x-directional velocity and temperature
- Contours of turbulent viscosity/diffusivity and turbulent Prandtl number
From additional analysis, it is discovered that local pressure gradient is highly related to the spatial variation of turbulent Prandtl number.
By modeling turbulent Prandtl number with appropriate modeling parameters, prediction error of turbulent heat transfer through RANS simulation is expected to be reduced significantly.
DNS for turbulent free jet & impinging jet
- Computational domain and jet simulations
- Vortical structures & contours of velocity of free jet
- Vortical structures & contours of velocity and temperature of impinging jet
- Contours of turbulent viscosity/diffusivity and turbulent Prandtl number of impinging jet
Numerical methods are validated through the velocity profiles of simulation of free jet.
Additional analysis will be performed with various approaches, such as turbulent budget terms and correlation coefficients.
DNS for turbulent flow in square duct
- Computational domain of square duct flow
- Vortical structures & contours of temperatures of duct flow
Validation of averaged velocity & vorticity fields are in progress