Forced flow of air over extended surfaces offers a simple, reliable, and effective heat removal mechanism and is often employed in electronic equipment. The IBM 4381 heat sink, used in production IBM computers, utilizes this cooling technique. This heat sink consists of a ceramic substrate on which fins made of an aluminum-copper alloy are arranged in a regular array. Cooling air enters the fin array from a nozzle. Extensive experiments have been carried out to characterize the performance of this heat sink at the Advanced Thermal Engineering Laboratory at IBM Endicott. This paper presents computational analysis of the three-dimensional flow and heat transfer in this device for two different air flow rates through the nozzle. The heat dissipated by the electronic components is conducted into the fins through the ceramic base. In the present study the ceramic base is assumed to be subjected to a uniform heat flux at the bottom. The computational method incorporates a special block-correction procedure to enable iterative solution of conjugate heat transfer in the presence of large differences in thermal conductivities of the air and the fin material. The results of computations reproduce the flow pattern in the fin array that is observed experimentally. The part of the ceramic base directly below the nozzle is well cooled with the temperatures gradually increasing from the center towards the corner. The predicted pressure drop and most of the local temperatures at the base and the tip of the fins agree well with the experimental observations. This study illustrates the utility of computational flow analysis in the analysis and design of electronic cooling techniques.

This paper proposes a magnetohydrodynamic model that does not assume local thermal equilibrium (LTE) to predict the flow field and electron temperature and species density distributions inside a direct-current nontransferred-arc steam torch with a well-type cathode at atmospheric pressure. A steady, axisymmetric flow is assumed, and the azimuthal velocity component is not neglected. A finite volume method is adopted to solve the continuity equation, the current continuity equation, the momentum equation, and the energy equation according to a k − ε turbulence model. Ohm's law is used to calculate the induced magnetic field, and Ampere's circuital law is adopted to compute the current distribution in space. Fifty-two chemical and plasma kinetic equations are employed to describe the interactions among the 12 main species of water plasma, H electrons, considered in this paper. The electron temperature is predicted from the transport equation based on the electron energy balance, whereas the transport coefficients of the plasma are obtained from the Chapman-Enskog solution of the Boltzmann equation. The thermal plasma flow of steam inside the plasma torch, which measures 503 mm in length and 9 mm in radius, is predicted to have I = 180 A and Q = 5 g/s. The non-LTE simulation suggests that a plasma temperature of 11 600 K and flow velocity of 3.8 km/s can be achieved at the center of the torch outlet. Compared with the LTE model, the non-LTE calculation yields lower plasma temperature and higher axial velocity at the torch outlet. The calculated electron temperature principally varies between 1 and 3 eV, and energetic electrons occur not only near the electrode surfaces but also within the axial electric arc. The model predictions suggest that the thermal plasma is well approximated as LTE in the hot plasma core, which is characterized by an isothermal contour of 10 kK, and in the vicinity of the electrodes. The deviation from the LTE condition predominantly grows with the radial coordinate, and the temperature ratio approaches unity at the electrode surfaces. The dominant species components of the water plasma in the inner region (r < 6 mm) at the torch outlet are monoatomic and ionic species of H and O, whereas the diatomic species and water molecule become relatively important in the outer region (r > 6 mm) at the torch outlet.

No abstract