Challenges in modeling of accelerated life testing

There exists a need to produce "real world" failures quickly for a product well before the product has been introduced into the field. This has lead to accelerated life testing (ALT), which details a procedure that attempts to mimic the product's behavior through a reduced number of tests, in a greatly reduced time period often in less than two weeks. The thermal performance is generally measured through many cycles of thermal shock or power variation in differing ambient environments. Capturing this behavior of entire ALT life through the use of simulation is combined with empirical relations to predict the life. In using simulation, the great concern is determining what level of detail needs to be included in the model to capture the relevant physics. With today's modern algorithms and vast computing power, simulations need not be simplistic models of the actual problem. However, one can see that this can lead to many challenges when looking at even the simplest packages. In addition, if the manufacturing variability (such as pad size and thickness, properties of the thin film/layers) need to be considered, the challenges are even greater. Since the material behavior varies as the temperature changes, additional modeling task are imposed.