Iztok Tiselj

Effect of wall boundary condition on scalar transfer in a fully developed turbulent flume

We performed direct numerical simulation of fully developed turbulent velocity and temperature fields in a flume, for Reynolds number, based on the wall shear velocity and the height of the flume, Re=171 and Prandtl numbers Pr=1.0 and Pr=5.4. To elucidate exactly the role of the wall boundary condition for passive scalar, the system considered was the flow at constant properties of the fluid. Two types of thermal wall boundary conditions (BCs) for the dimensionless temperature equation were studied: isothermal wall boundary condition—H1, and isoflux wall boundary condition—H2. The profile of the mean temperature was not affected by the type of BC. However, the type of BC has a profound effect on the statistics of the temperature fluctuations in the near-wall region y+<10. Comparison of near-wall statistics of temperature fluctuations shows that at Pr=1 the buffer part of the turbulent boundary layer significantly influences the scalar transfer in the conductive sublayer, whereas at Pr=5.4 the near-wall te...

DNS of Turbulent Heat Transfer in Channel Flow With Heat Conduction in the Solid Wall

The Direct Numerical Simulation (DNS) of the fully developed velocity and temperature fields in the two-dimensional turbulent channel flow coupled with the unsteady conduction in the heated walls was carried out. Simulations were performed at constant friction Reynolds number 150 and Prandtl numbers between 0.71 and 7 considering the fluid temperature as a passive scalar. The obtained statistical quantities like root-mean-square temperature fluctuations and turbulent heat fluxes were verified with existing DNS studies obtained with ideal thermal boundary conditions. Results of the present study were compared to the findings of Polyakov (1974), who made a similar study with linearization of the fluid equations in the viscous sublayer that allowed analytical approach and results of Kasagi et al. (1989), who performed similar calculation with deterministic near-wall turbulence model and numerical approach. The present DNS results pointed to the main weakness of the previous studies, which underestimated the values of the wall temperature fluctuations for the limiting cases of the ideal-isoflux boundary conditions. With the results of the present DNS it can be decided, which behavior has to be expected in a real fluid-solid system and which one of the limiting boundary conditions is valid for calculation, or whether more expensive conjugate heat transfer calculation is required. @DOI: 10.1115/1.1389060#

Near-wall passive scalar transport at high Prandtl numbers

Very accurate numerical simulations of a passive scalar field in the turbulent channel and flume flow were performed at friction Reynolds numbers Reτ=150 and Reτ=395 and Prandtl numbers Pr=100, Pr=200. Direct numerical simulation is used for description of the velocity field. The temperature field is described with the LES-like approach with the smallest resolved temperature scales equal to the smallest scales of the velocity field. The consistency of the applied physical modelling and pseudospectral scheme is first tested with comparison of the results with the existing DNS simulations of F. Schwertfirm and M. Manhart [Proceedings of Turbulence, Heat, and Mass Transfer (2006)] at Reτ=180 and Pr=25. The sensitivity of the method to the grid refinement and time step variations is performed with simulations at Reτ=150 and Pr=200. Both tests show that the proposed approach produces very accurate mean temperature profiles, heat transfer coefficients, and other low-order moments of the turbulent thermal field....

Some comments on the behavior of the RELAP5 numerical scheme at very small time steps

Abstract The behavior of the RELAP5 code at very short time steps is described, i.e., Δt ≈ 0.01 Δx/c. First, the property of the RELAP5 code to trace acoustic waves with “almost” second-order accuracy is demonstrated. Quasi-second-order accuracy is usually achieved for acoustic waves at very short time steps but can never be achieved for the propagation of nonacoustic temperature and void fraction waves. While this feature may be beneficial for the simulations of fast transients describing pressure waves, it also has an adverse effect: The lack of numerical diffusion at very short time steps can cause typical second-order numerical oscillations near steep pressure jumps. This behavior explains why an automatic halving of the time step, which is used in RELAP5 when numerical difficulties are encountered, in some cases leads to the failure of the simulation. Second, the integration of the stiff interphase exchange terms in RELAP5 is studied. For transients with flashing and/or rapid condensation as the main phenomena, results strongly depend on the time step used. Poor accuracy is achieved with “normal” time steps (Δt ≈ Δx/v) because of the very short characteristic timescale of the interphase mass and heat transfer sources. In such cases significantly different results are predicted with very short time steps because of the more accurate integration of the stiff interphase exchange terms.

Heat transfer and thermal pattern around a sphere in a turbulent boundary layer

Abstract Direct numerical simulation (DNS) is performed by solving the governing equations for fluid flow and heat transfer. The nondimensional sphere diameters are 17 and 34 wall units and cover several collocation points of the fluid. Two-way coupling is used to account for the effect of the sphere on the structure of the near-wall turbulence and the main stream. The calculation of the thermal field is done with the same grid system used for the velocity field. Water was used as test fluid, with the Prandtl number Pr =5.4. The calculation is performed for a single stationary sphere attached to the bottom of the flume. The heat transfer calculations were carried out at a constant heat flux wall boundary condition. The present DNS results indicate essential enhancement of heat transfer coefficient associated with a flow motion toward the wall. The thermal pattern around the sphere is obtained and compared with the experimental images. A possible mechanism of heat transfer in the presence of coarse particles in the near-wall region of a turbulent boundary layer is discussed.

WALL PROPERTIES AND HEAT TRANSFER IN NEAR-WALL TURBULENT FLOW

Direct numerical simulation of a passive scalar in fully developed turbulent channel flow is used to show that Nusselt number is not only a function of Reynolds and Prandtl number, but also depends on properties of a heating wall. Variable thickness of the heating wall and variable heater properties, combined in a fluid–solid thermal activity ratio , can change the Nusselt number of the turbulent channel flow for up to 1% at the same Reynolds and Prandtl number and at the same wall heat flux.

First and Second-Order Accurate Schemes for Two-Fluid Models

The six-equation two-fluid model of two-phase flow taken from the RELAP5/MOD3 computer code has been used to simulate three simple transients: a two-phase shock tube problem, the Edwards Pipe experiment, and water hammer due to rapid valve closure. These transients can be characterized as fast transients, since their characteristic time-scales are determined by the sonic velocity. First and second-order accurate numerical methods have been applied both based on the well-known, Godunov-type numerical schemes. Regarding the uncertainty of the two-fluid models in todays large computer codes for the nuclear thermal-hydraulics, use of second-order schemes is not always justified. While this paper shows the obvious advantage of second-order schemes in the area of fast transients, first-order accurate schemes may still be sufficient for a wide range of two-phase flow transients where the convection terms play a minor role compared to the source terms

Accuracy of the Operator Splitting Technique for Two-Phase Flow With Stiff Source Terms

Code for analysis of the water hammer in thermal-hydraulic systems is being developed within the WAHALoads project founded by the European Commission [1]. Code will be specialized for the simulations of the two-phase water hammer phenomena with the two-fluid model of two-phase flow. The proposed numerical scheme is a two-step second-order accurate scheme with operator splitting; i.e. convection and sources are treated separately. Operator splitting technique is a very simple and “easy-to-use” tool, however, when the source terms are stiff, operator splitting method becomes a source of a specific non-accuracy, which behaves as a numerical diffusion. This type of error is analyzed in the present paper.Copyright

Convection Velocity of Temperature Fluctuations in a Turbulent Flume

A numerical investigation of the temperature field in a turbulent flume is presented. We consider the effect of the Prandtl number on the convection velocity of temperature fluctuations in a turbulent boundary layer, and focus also on the effect of the Prandtl number on the connection between the velocity and the temperature fluctuations

Tracking of large-scale structures in turbulent channel with direct numerical simulation of low Prandtl number passive scalar

Channel flow DNS (Direct Numerical Simulation) at friction Reynolds number 180 and with passive scalars of Prandtl numbers 1 and 0.01 was performed in various computational domains. The “normal” size domain was ∼2300 wall units long and ∼750 wall units wide; size taken from the similar DNS of Moser et al. The “large” computational domain, which is supposed to be sufficient to describe the largest structures of the turbulent flows was 3 times longer and 3 times wider than the “normal” domain. The “very large” domain was 6 times longer and 6 times wider than the “normal” domain. All simulations were performed with the same spatial and temporal resolution. Comparison of the standard and large computational domains shows the velocity field statistics (mean velocity, root-mean-square (RMS) fluctuations, and turbulent Reynolds stresses) that are within 1%-2%. Similar agreement is observed for Pr = 1 temperature fields and can be observed also for the mean temperature profiles at Pr = 0.01. These differences can...