Issue link: https://iconnect007.uberflip.com/i/424967
36 SMT Magazine • December 2014 the second. Were it possible, optimizing both steps in one would effectively create the flex- ibility needed for the factory, while retaining productivity as near to that of running high volume. A subtle change in the way that we ap- proach planning optimization in the future will have a profound effect on the operation of the factory. We need to look at each of the elements to understand the elements involved in the op- timization of SMT production. From the SMT- machine-based tool perspective, tra- ditional SMT optimization relates to the machine environment, making sure that the machines and lines are themselves fully optimized. At the base level, each SMT machine program is optimized to ensure that the machine is adding value for as much of the operating time as possible, without needless ac- tions that slow performance. To put just this level of optimi- zation into perspective, if ev- ery possible sequence of SMT placement path were consid- ered by a super-computer, it would take years of process- ing to find the best sequence. SMT optimization algorithms in themselves can be software "works of art," designed to avoid having to consider every possible permutation. On top of this optimization, however, the overall work for each PCB must usually be split between the different SMT ma- chines and other processes in the line. Line optimization ensures that the best machines are used for each material, and, that the execu- tion times between machines in the line will end up being the same, balanced; after all, the line is only as fast as the slowest process. Go- ing up another level in the SMT level optimiza- tion, there is then the consideration of the line down-time that is needed to change material setups at the machine between one product and the next. This can represent more operational loss time than any other optimization factor in higher mix manufacturing. This optimiza- tion is achieved through the commonisation of materials feeder setup positions across a group of different products that will run sequentially. The machine-level software will create the setup based on a given list of products. These elements of SMT optimization can together be thought of as the "three dimen- sions of SMT optimization," and are supported by the best SMT programming tools currently available, either from machine vendors directly where only one vendor of machines is used in a line, or by third- party multi-vendor solutions. The weakness of these tools and the inherent barrier for them to overcome is that the choice of products with which to make the common material setups is left as an open input, not considered for optimization, and is usu- ally taken from the output of the shop-floor planning sys- tem. Unfortunately, the shop- floor planning system also has the same weakness in that it is not able to understand or consider the SMT optimiza- tion criteria. It can only opti- mize based on standard plan- ning criteria, which includes delivery requirements, materi- als and other resource availabil- ity, asset performance, and current shop-floor progress. Ideally, there should be one system to optimize the shop-floor planning together with the SMT optimization—the 4 th di- mension of SMT shop-floor optimization. Even with the smartest of algorithms, however, opti- mization of all four dimensions of this problem simultaneously is nowhere near (yet) the scope of today's computing power, so what can we do? The responsive Factory of the Future The solution to our planning conundrum is to turn the problem on its head. Instead of the serious compromise being made by first opti- THE ESSEnTiAl pionEEr'S SurvivAl guiDE THe FUTUre OF SmT: WeLcOme TO THe 4 TH dImeNSION continues Ideally, there should be one system to optimize the shop-floor planning together with the SMT optimization—the 4 th dimension of SMT shop-floor optimization. Even with the smartest of algorithms, however, optimization of all four dimensions of this problem simultaneously is nowhere near (yet) the scope of today's computing power, so what can we do? " "