Tsuguo Kondoh

Computational study of laminar heat transfer downstream of a backward-facing step

Abstract Laminar heat transfer in a separating and reattaching flow was numerically studied by simulating the flow and heat transfer downstream of a backward-facing step. A series of computations was conducted in which three principal parameters governing the heat transfer in this geometry (i.e. channel expansion ratio ER, Reynolds number Re and Prandtl number Pr) were systematically changed. As a result, detailed relations between these parameters and the fundamental heat transfer characteristics have been elucidated. Some important findings are: (1) the distribution of the local Nusselt number depends strongly on all of these parameters; (2) the peak of the local Nusselt number does not necessarily locate at or very near the point of flow reattachment, in contrast to the common belief; and (3) if the Prandtl number is considerably low, the peak itself does not even appear and hence the heat transfer enhancement, usually assumed around the flow reattachment point, can never be expected in such cases.

A Mixed-Time-Scale SGS Model With Fixed Model-Parameters for Practical LES

ABSTRACT A new subgrid-scale (SGS) model for practical large eddy simulation (LES) is proposed. The model is constructed with the concept of mixed (or hybrid) time scale, which makes it possible to use consistent model parameters and to dispense with the distance from the wall. The model performance is tested in plane channel flows, and the results show that this model is able to account for near-wall turbulence without an explicit damping function as in the dynamic Smagorinsky model. The model is also applied to the backward-facing step flow examined by Kasagi and Matsunaga (1995) experimentally. The calculated results show good agreement with experimental data, while the results obtained using the dynamic Smagorinsky model show less accuracy and less computational stability. To confirm the validity of the present model in practical applications, the three-dimensional complex flow around a bluff body ( Ahmed, 1984 ) is also calculated with the model. The agreement between the calculated results and the experimental data is quite satisfactory. These results suggest that the present model is a refined SGS model suited for practical LES to compute flows in a complicated geometry.

On Reynolds-stress expressions and near-wall scaling parameters for predicting wall and homogeneous turbulent shear flows

Abstract In this paper, a new type of two-equation turbulence model that incorporates some essential characteristics of second-order closure models is proposed. The present model belongs to a nonlinear k−ϵ model taking into account low-Reynolds-number effects originating from the physical requirements, and is applicable to complex turbulent flows with separation and reattachment. The model successfully predicts both wall-turbulent and homogeneous shear flows, the latter of which has been very difficult to simulate with existing two-equation turbulence models. Channel flows with injection and suction at wall surfaces and separated and reattaching flows downstream of a backward-facing step are also calculated. Comparisons of the computational results with the measurements and the direct numerical simulation data indicate that the present model is effective in calculating complex turbulent flows of technological interest. Furthermore, the parameters for scaling the near-wall region in the low-Reynolds-number model functions are re-evaluated, yielding some insights into the near-wall scaling parameters for application to complex turbulent flows.

A two-equation heat transfer model reflecting second-moment closures for wall and free turbulent flows

Abstract A new two-equation heat transfer model which incorporates some essential features of second-order modeling is proposed. The present model is applicable to heat transfer problems in both wall and free turbulent flows not unusually deviating from the equilibrium state. Furthermore, by introducing the Kolmogorov velocity scale, the model can appropriately express the low Reynolds number effects in the near-wall region and is also applicable to complex heat transfer fields with flow separation and reattachment. It is shown that the proposed model predicts quite successfully heat transfer in both wall and free turbulent flows; i.e., a homogeneous isotropic decaying flow, a homogeneous shear flow, a boundary-layer flow heated from the origin, and a boundary-layer flow subjected to a sudden change in the wall-heating condition; whereas, such predictions have been almost impossible with existing two-equation heat transfer models.

Numerical prediction of fluid-resonant oscillation at low Mach number

A method (governing equation set and numerical procedure) suited to the numerical simulation of fluid-resonant oscillation at low Mach numbers is constructed. The new equation set has been derived under the assumption that the compressibility effect is weak. Because the derived equations are essentially the same as the incompressible Navier-Stokes equations, except for an additional term, we can apply almost the same numerical procedure developed for incompressible flow equations without difficulty. With application of a pressure-based method that treats the continuity equation as a constraint equation for pressure, the stiffness problem that arises in solving the usual compressible flow equations under low Mach number conditions has been alleviated. To verify the present method, we apply it to the flows over a three-dimensional open cavity

Numerical prediction of separating and reattaching flows with a modified low-Reynolds-number k-ϵ model

We propose a new turbulence model which is modified from the low-Reynolds-number k-ϵ model developed by Nagano and Tagawa. The main improvement is made by the introduction of the Kolmogorov velocity scale uϵ ≡ (νϵ)14 instead of the friction velocity uτ. We have also reevaluated the model constants in the transport equations. With these modifications, it is shown that the present model predicts quite successfully the separating and reattaching turbulent flows downstream of a backward-facing step at various Reynolds numbers, which have been very difficult to simulate with any previous k-ϵ model.

A MIXED-TIME-SCALE SGS MODEL WITH FIXED MODEL-PARAMETERS FOR PRACTICAL LES

A new subgrid-scale (SGS) model for practical large eddy simulation (LES) is proposed. The model is constructed with the concept of mixed (or hybrid) time scale, which makes it possible to use consistent model parameters and to dispense with the distance from the wall. The model performance is tested in plane channel flows, and the results show that this model is able to account for near-wall turbulence without an explicit damping function as in the dynamic Smagorinsky model. The model is also applied to the backward-facing step flow examined by Kasagi and Matsunaga (1995) experimentally. The calculated results show good agreement with experimental data, while the results obtained using the dynamic Smagorinsky model show less accuracy and less computational stability. To confirm the validity of the present model in practical applications, the three-dimensional complex flow around a bluff body (Ahmed, 1984) is also calculated with the model. The agreement between the calculated results and the experimental data is quite satisfactory. These results suggest that the present model is a refined SGS model suited for practical LES to compute flows in a complicated geometry.

Topology Optimization of Power Semiconductor Devices

In this paper, topology optimization is applied to the design of power semiconductor devices. The doping density distribution of power semiconductor devices is optimized using a density-based topology optimization method. The density method is suitable for the design of power semiconductor devices because doping density can take on continuous values and is intrinsically free from the gray-scale problem. To verify the effectiveness of topology optimization, optimization was conducted for two types of two-dimensional design problems. At first, optimization of a p-n diode was performed to improve the trade-off between the breakdown voltage and on-resistance. This is formulated as a single-objective optimization problem with the Kreisselmeier-Steinhauser (KS) objective function of the electric field, which indicates the breakdown voltage characteristics, under the constraint of the on-resistance. By optimization, a p-i-n diode, which is a well-known diode structure, is obtained and the trade-off is improved. Next, optimization of an edge termination structure was performed to improve the breakdown voltage characteristics with the consideration of ion implantation, which is one of the fabrication processes used for semiconductor devices, under the process variation. The optimized structure obtained is ensured to be manufacturable and more robust with respect to the dose amount variation of ion implantation than the initial structure. These results demonstrate the effectiveness of topology optimization for the design of power semiconductor devices.