Engineering topics

Nakamura, S.

Publications and source records attributed to Nakamura, S..

11 records

NASA NTRS1999

The Role of Marangoni Convection for the FZ-Growth of Silicon

Growing crystals by the floating zone (FZ) technique under microgravity avoids the size restriction we have under earth conditions due to hydrostatic pressure. Further, buoyancy related convection is eliminated to a great degree. But in the case of silicon, the gravity independent thermocapillary (Marangoni) convection is time-dependent even for small zone geometries. This has been demonstrated in several Technische Experimente unter Schwerelosigkeit (TEXUS) technical experiments under reduced gravity flights. Thus, to really take advantage of microgravity with respect to improved crystal quality, tools are required to control Marangoni convection in space facilities. With the application of magnetic fields, convection can be influenced; fluid flow can either be damped (static magnetic fields) or overlaid by a regular flow regime (rotating magnetic fields). In floating zones of 8-10 mm diameter and height (i.e., Ma is much greater than 6X 10(exp 3), a static magnetic field of about 2OOmT is sufficient to suppress time-dependent Marangoni convection to a high degree, but in dependence on the kind and the concentration of the added dopant, a new type of strongly pronounced dopant inhomogeneities have been detected. They are originated by thermoelectromagnetic convection. This can be avoided as well as detrimental effects on the radial dopant distribution by using rotating magnetic fields instead of static ones. Applying 7.5mT/5OHz to the FZ, the intensity of the dopant fluctuations is reduced to a high degree. Considering the rather low power consumption of rotating magnetic fields, this will be a useful tool for control or elimination of time-dependent Marangoni convection under microgravity. The strong time-dependent character of thermocapillary flow and its influence on the temperature field has been measured in silicon half-zones for Marangoni numbers of Ma is much greater than l - 1.5 X 10(exp 4): temperature fluctuations up to 4C have been determined, their frequency range was 0.4 and 0.4Hz. Between certain thermocouple or sensor pairs, strong correlation has been detected.

Dold, P.↗

NASA NTRS1998

The Role of Marangoni Convection for the FZ-Growth of Silicon

Fluctuations of the electrical resistivity due to inhomogeneous dopant distribution are still a serious problem for the industrial processing yield of doped silicon crystals. In the case of silicon floating-zone growth, the main sources of these inhomogeneities are time- dependent flows in the liquid phase during the growth process. Excluding radio frequency (RF) induced convection, buoyancy and thermocapillary (Marangoni) convection are the two natural reasons for fluid flow. Both originate from temperature/concentration gradients in the melt, buoyancy convection through thermal/concentrational volume expansion, and thermocapillary convection through the temperature/concentration dependence of the surface tension. To improve the properties of grown crystals, knowledge of the strength, the characteristic, and the relation of these two flow mechanisms is essential. By the use of microgravity, the effect and the strength of buoyancy (gravity dependent) and thermocapillary (gravity independent) convection can be separated and clarified. Applying magnetic fields, both convective modes can be influenced: fluid flow can either be damped (static magnetic fields) or overlaid by a regular flow regime (rotating magnetic fields). Two complementary approaches have been pursued: Silicon full zones (experiments on the German sounding rockets TEXUS 7, 12, 22, 29, and 36) with the maximum temperature at half of the zone height and silicon half zones (experiments on the Japanese sounding rockets TR-IA4 and 6) with the maximum temperature at the top of the melt. With the full zone arrangement, the intensity and the frequency of the dopant striations could be determined and the critical Marangoni number could be identified. The half zone configuration is suited to classify the flow pattern and to measure the amplitude and the frequency of temperature fluctuations in the melt by inserting thermocouples or temperature sensors into the melt. All experiments have been carried out in monoellipsoid mirror furnaces. Typical zone geometries are approx. 8 to 14 mm in diameter and height. The crystals grown under microgravity are compared to crystals grown in static axial magnetic fields (B<5 tesla) and in transversal rotating magnetic fields (B<7.5 mT / f=50 Hz). Experimental results are completed by 3D numerical simulations: the obtained temperature and concentration distribution in the melt confirm the damping effect of rotating magnetic fields and explain the change in the radial segregation under static magnetic fields.

Dold, P.↗

NASA NTRS1998

The Role of Marangoni Convection for the FZ-Growth of Silicon

In growing crystals by the floating zone (FZ) technique under microgravity, the size restriction we have under earth conditions because of the hydrostatic pressure are avoided. Further, buoyancy related convection is eliminated to a high degree. But in the case of silicon, the gravity independent thermocapillary (Marangoni) convection is time-dependent even for small zone geometries. This has been demonstrated in several Technische Experimente unter Schwerelosigkeit (TEXUS) - technical experiments under reduced gravity) flights. Thus, to really take advantage of microgravity with respect to improve crystal quality, tools are required to control Marangoni convection in space facilities. Applying magnetic fields, convection can be influenced; fluid flow can either be damped (static magnetic fields) or overlaid by a regular flow regime (rotating magnetic fields). In floating zones of 8-10mm diameter and height (Ma approximately equals 6 x 10(exp 3)), a static magnetic field of about 2OOmT is sufficient to suppress time-dependent Marangoni convection to a high degree, but in dependence on the kind and the concentration of the added dopant, a new type of strongly pronounced dopant inhomogeneities have been detected. They are originated by thermoelectromagnetic convection. This can be avoided as well as detrimental effects on the radial dopant distribution by using rotating magnetic fields instead of static ones. Applying 75mT/50Hz to the FZ, the intensity of the dopant fluctuations is reduced to a high degree. Considering the rather low power consumption of rotating magnetic fields, this will be a useful tool for control or elimination of time-dependent Marangoni convection under microgravity. The strong time dependent character of thermocapillary flow and its influence on the temperature field has been measured in silicon half-zones for Marangoni numbers of Ma is approximately equal to 1-1.5 x 10(exp 4): temperature fluctuations up to 4C have been determined. Their frequency range was 0.1 and 0.4 Hz. Between certain thermocouple or sensor pairs, strong correlation has been detected.

Dold, P.↗

NASA NTRS1993

Simulation of unsteady flows through stator and rotor blades of a gas turbine using the Chimera method

A two-dimensional model to solve compressible Navier-Stokes equations for the flow through stator and rotor blades of a turbine is developed. The flow domains for the stator and rotor blades are coupled by the Chimera method that makes grid generation easy and enhances accuracy because the area of the grid that have high turning of grid lines or high skewness can be eliminated from the computational domain after the grids are generated. The results of flow computations show various important features of unsteady flows including the acoustic waves interacting with boundary layers, Karman vortex shedding from the trailing edge of the stator blades, pulsating incoming flow to a rotor blade from passing stator blades, and flow separation from both suction and pressure sides of the rotor blades.

Nakamura, S.↗

NASA NTRS1992

Grid generation research at OSU

In the last two years, effort was concentrated on: (1) surface modeling; (2) surface grid generation; and (3) 3-D flow space grid generation. The surface modeling shares the same objectives as the surface modeling in computer aided design (CAD), so software available in CAD can in principle be used for solid modeling. Unfortunately, however, the CAD software cannot be easily used in practice for grid generation purposes, because they are not designed to provide appropriate data base for grid generation. Therefore, we started developing a generalized surface modeling software from scratch, that provides the data base for the surface grid generation. Generating surface grid is an important step in generating a 3-D space for flow space. To generate a surface grid on a given surface representation, we developed a unique algorithm that works on any non-smooth surfaces. Once the surface grid is generated, a 3-D space can be generated. For this purpose, we also developed a new algorithm, which is a hybrid of the hyperbolic and the elliptic grid generation methods. With this hybrid method, orthogonality of the grid near the solid boundary can be easily achieved without introducing empirical fudge factors. Work to develop 2-D and 3-D grids for turbomachinery blade geometries was performed, and as an extension of this research we are planning to develop an adaptive grid procedure with an interactive grid environment.

Nakamura, S.↗

NASA NTRS1983

Generation of orthogonal grids by boundary grid relaxation

Two elliptic grid generation schemes that yield orthogonal grids are presented (FGBR and UBGR). With the UBGR scheme, the grid points on the flow boundaries are automatically determined by the algorithm, while the FBGR scheme at most one half of the boundary grid points may be prespecified and the remainder of boundary grid points are determined automatically. Numerical examples show their capability of easy stretching, clustering and shock fitting while maintaining orthogonality of grid. The present method can be implemented into existing elliptic grid generation programs with relatively minor modifications.

Nakamura, S.↗

NASA NTRS1983

Evaluation of truncation error and adaptive grid generation for the transonic full potential flow calculations

The effects of truncation error on the numerical solution of transonic flows using the full potential equation are studied. The effects of adapting grid point distributions to various solution aspects including shock waves is also discussed. A conclusion is that a rapid change of grid spacing is damaging to the accuracy of the flow solution. Therefore, in a solution adaptive grid application an optimal grid is obtained as a tradeoff between the amount of grid refinement and the rate of grid stretching.

Nakamura, S.↗

NASA NTRS1983

Adaptive grid relocation algorithms for transonic full potential calculations using one-dimensional or two-dimensional diffusion equation

The effect of solution-adaptive grids using an interpolative grid relocation algorithm on two-dimensional transonic full potential flow calculations is studied with extensions to three dimensions in mind. One-dimensional or two-dimensional diffusion equation is solved to obtain the grid density control function for generating adaptive grids. The results of flow solution for NACA-0012 and Gates-Leajet airfoils show a significant improvement of accuracy in the CP distributions. However, further studies are recommended on the questions (1) what information from the initial flow solution should be passed on to the solution-adaptive grid generation scheme, and (2) how to optimize the accuracy of the solution through the use of adaptive grids.

Nakamura, S.↗

NASA NTRS1982

Marching grid generation using parabolic partial differential equations

The feasibility of using parabolic partial differential equations for grid generation is examined. Source terms in the form of a linear interpolation between the current grid and the outer boundary are used in generating a grid for a two-dimensional airfoil flow. Grid generation equations are derived by assuming that grid spacings are locally nonuniform on the computational domain. The grid spacing control method is described in detail, and the local orthogonality of the grid lines is discussed. An O-mesh and an H-mesh generated by the described method are shown.

Nakamura, S.↗

NASA NTRS1982

Noniterative grid generation using parabolic difference equations for fuselage-wing flow calculations

A fast method for generating three-dimensional grids for fuselage-wing transonic flow calculations using parabolic difference equations is described. No iterative scheme is used in the three-dimensional sense; grids are generated from one grid surface to the next starting from the fuselage surface. The computational procedure is similar to the iterative solution of the two-dimensional heat conduction equation. The proposed method is at least 10 times faster than the elliptic grid generation method and has much smaller memory requirements. Results are presented for a fuselage and wing of NACA-0012 section and thickness ratio of 10 percent. Although only H-grids are demonstrated, the present technique should be applicable to C-grids and O-grids in three dimensions.

Nakamura, S.↗

NASA NTRS1981

A new solution-adaptive grid generation method for transonic airfoil flow calculations

The clustering algorithm is controlled by a second-order, ordinary differential equation which uses the airfoil surface density gradient as a forcing function. The solution to this differential equation produces a surface grid distribution which is automatically clustered in regions with large gradients. The interior grid points are established from this surface distribution by using an interpolation scheme which is fast and retains the desirable properties of the original grid generated from the standard elliptic equation approach.

Nakamura, S.↗