Résumé:
The rapid development of technologies related to microprocessors, already achieves several
GHz, while reducing the size of microchips. Therefore, the flux density of heat generated by
these circuits is constantly increasing, this gives birth to their heating. One major problem lies
in the disposal and transport of the energy dissipated by these systems. This trend of
miniaturization has serious operational constraints for these components, particularly at the
operating temperature. This thesis is therefore a logical consequence of simplifying and
improving the thermal management of power electronic components. The objective of this
study consists of thermal and hydrodynamic modeling of a cooling system for
microprocessor. The system is a heat sink of plate fins is subjected to an air jet impingement.
In order to analyze the parameters those control the cooling mechanism for possible
improvement. To do this, the thermal and hydrodynamic modeling and boundary conditions
imposed on the system are presented. The resolution of governing equations is then performed
by Fluent CFD code and a developed code based on the finite difference method. This study
has clearly identified the advantages of numerical tools to reflect and reproduce the physical
phenomena occurring in the heat sink. After securing the validation step, the code has been
used to study the effect of different geometric parameters and dynamic characteristics of flow
and therefore on improving the heat transfer. Predicting the performance of hydrodynamic
and thermal cooling device shows that the microprocessor to deliver up to 80W of power is
properly cooled by the heat sink as indicated by the manufacturer. We noted the importance
of geometric and dynamic characteristics in enhancing the capacity cooling of the heat sink.
For example, overall improvements in the thermal resistance of the radiator of ~ 7% which
can occur by increasing the cooling rate of 5 m/s to 6m/s only. Therefore, the thermohydrodynamic
performance of the heat sink can be further improved, hence the use of this
simple and economic technique remains a preferred solution for thermal management of
power electronics components.
