Enhancement of the survival probability of a photovoltaic converter-An optimization approach

As a first step, the converter reliability is estimated according to the FIDES procedure, yielding a mean time between failures MTBF = 98.15 FIT. The numerical results pinpoint the diodes as the most failure-prone components, and temperature as the dominant stressor, clearly suggesting that the performance can be improved by resorting to better heatsinks. As a second step, a heatsink optimization problem is established, involving three objective functions: minimization of heatsink thermal resistance and weight, and maximization of the cooling system performance index, The problem is solved using a genetic algorithm, which yields MTBF = 74.09 FIT. The approach takes advantage of the computer tools currently available: mathematical toolboxes to implement the optimization algorithm, user-friendly reliability prediction software, and software packages capable of simulating the thermal behavior of heatsinks with good accuracy.