Estimating and Improving the Production Rate and Machine Utilization of an Automated Component Insertion Line - Part II.

Many shops such as the D5 line in Merrimack Valley, the AIM line in North Carolina and the CIM lines in Oklahoma City assemble circuit packs using highly automated insertion equipment and the inline manufacturing concept. More new shops, the DDM1000 line in North Carolina for example, are being built. In these lines, relatively small buffers are assigned between workstations in order to guarantee a small manufacturing interval and quick feedback in the event of process failure to make an acceptable product. These small buffers could lead to a noticeable loss of insertion capacity due to the phenomenon of blocking and starvation. In this memorandum, we develop two fast iterative algorithms to estimate the insertion capacity and machine utilizations for a given configuration of machines and buffer sizes and for a given code mix and loading sequence. These algorithms are based on the LP method developed in Part I. The first algorithm provides an upper bound while the second a pessimistic figure for the insertion rate. A small (large) difference between these two figures generally indicates that simple scheduling will (will not) suffice to obtain the performance predicted by these algorithms. Some real data from the DDM1000 line is used to illustrate a heuristic method for improving the performance of a line. In this method, these algorithms are run repeatedly and after each run one or more of the following steps are taken: (a) make codes in small or larger lots, (b) alter the sequence of production, (c) increase the buffer space selectively and (d) add new machines. This is the first method of its type which, in addition to evaluating a given production scenario, also provides insights and guidance as to how improvements can be made. This provides an important tool for designing new lines and for operating the existing ones more efficiently.