The demand for high execution speed and memory capacity for modern computers results in an increasing circuit density per unit chip and high power dissipation per unit volume. Consequently, traditional air cooling technology such as air-cooled heat sink is reaching the limits for electronic applications. Thermoelectric coolers are regarded as potential solutions for enhancing the performance of air-cooled heat sinks. In the present study, a semi-empirical method for exploring the thermal performance of a heat sink integrated with or without TEC has been successfully established. A concept of design of experiments (DOE) is applied, and a statistical method for sensitivity analysis of the influencing parameters is performed to determine the key factors that are critical to the design. By the statistical sensitivity analysis of ANOVA F-test for the temperature reduction (ΔTC−B) and COP of the TEC, the factor contributions of QP, Rext and I are 31.66%, 33.73%, 34.61% as well as 14.9%, 0%, 85.1%, respectively. By employing the gradient-based numerical optimization technique, a series of constrained optimal designs have been performed. Under the given constraints of COP≧2, the optimal value of ΔTC−B (3.3°C) is obtained with the corresponding Qp (31.99W) and Qte (16W). Comparisons between the results by the present optimal design and those obtained by the semi-empirical results have been made with a satisfactory agreement. The present optimal design shows that a heat sink integrated with TEC can extend the upper limits of thermal management for traditional air-cooled heat sinks.
How Good Is the STW Sensor? An Account from a Larger Shipping Company