Strategies to the Prediction, Mitigation and Management of Product Obsolescence

4.2 OBSOLESCENCE FORECASTING—PARTS WITHOUT EVOLUTIONARY PARAMETRIC DRIVERS (SANDBORN ET AL., 2011)

This section introduces a methodology for generating algorithms that can be used to predict the obsolescence dates for electronic parts that do not have clear evolutionary parametric drivers. The method is based on the calculation of procurement lifetime using databases of previous obsolescence events and introduced parts that have not gone obsolete. The methodology is demonstrated on a range of different discrete semiconductor electronic parts for the trending of specific part attributes.

4.2.1 Procurement Lifetime

The previously discussed data mining method for forecasting obsolescence has been shown to work well when there are identifiable evolutionary parametric drivers (Section 4.1). An evolutionary parametric driver is a parameter (or a combination of parameters) describing how the part evolves over time. For example, for flash memory chips an evolutionary parametric driver is memory size; traditionally for microprocessors it has been clock frequency (although recently this has begun to give way to power consumption). Unfortunately, for the majority of electronic parts, there is no simple evolutionary parametric driver that can be identified, and previously proposed data mining approaches cannot be used (diodes, transistors, operational amplifiers, multiplexer, buffers, and so on).

In this section, a methodology for formulating obsolescence forecasting algorithms based on predicting ...