Agile Manufacturing Systems

As packaging and test technology becomes more complex, especially with the advent of systems on a package enabled by silicon interposers (passive or active) there is a glaring gap in planning and decision systems tied to technology development and high volume manufacturing as will be outlined below. We will first start with the technology development problem statement and then expand this out to the high-volume network. Technology Development: In the development (TD) factory one is faced with a complex dynamic environment. The complexity comes from the high mix of test vehicles, prototypes, and engineering material, along with a steady flow of revenue products. The dynamism comes from the small volumes and one of a kind or first of a kind processes, equipment and collaterals which rely on an immature supply chain. The small volume material is quite often of very high strategic value as it could be a test build of a specific revision of a product, or samples being sent for reliability assessment. At the same time the small volume means it is often build to collateral availability – e.g. the correct number of silicon die arriving at the factory at the correct time – and any perturbation in collateral schedules, or process or equipment availability will disrupt the flow of this material. It is of interest of have close to real-time decision systems that will help plan the factory operational strategy, and enable rapid responses to the kinds of excursions mentioned above. It has been found that typical solvers for example (using constraint programming) take 30+ minutes of reasonable fast hardware, and this is deemed a slow response. Ideally the response needs to be determined in close to real-time. Note that the factory process flows represent a multi-echelon job shop type flow with setup times where some products may share a specific machine group. Figure 1 shows an example flow to illustrate the problem – this example shows 3 different types of products and the