Issue link: https://iconnect007.uberflip.com/i/1411055
SEPTEMBER 2021 I PCB007 MAGAZINE 13 of processes. Put the required date on individ- ual items at the last possible moment, i.e., fore- cast over the shortest possible horizon. Nev- er put into a manufacturing facility more than you believe can be produced. Input/Output Control: A Case Study In 1996, our production control manager wrote a paper for the California Circuits Asso- ciation Conference (1998) on the project to re- duce our lead time and lot size for multilayer production [5] . is is the only paper I can find about a lot or lead time case study for PCB fabrication. is paper will be available in the I-Connect007 Technical Library. Background In the late '90s, Hewlett-Packard's PCB fabrication facility found itself with increasing orders and increasing lead-time. e produc- tion control manager, Bill Nordskog, took action and this is an account of his efforts to control its lead-time and throughput. Motivation for Change It was clear that lead time was the princi- pal culprit creating our dilemma. To reduce the magnitude and duration of future de- mand fluctuations, we were certain that a sig- nificant lead time reduction was the required first step. A second motivating factor was the empha- sis top management had placed on improv- ing inventory and accounts receivable control following the analysis that a disproportionate increase in the levels of these two assets was forcing the company to consider significant long-term debt financing for the first time. By implementing a reduction in our work-in-pro- cess, we could make a positive contribution to avoid that prospective debt. Finally, recalling Oliver Wight's fine arti- cle, "Input/Output Control: A Real Handle on Lead Time," [1] indicated to us that many other problems our shop was facing (late de- liveries, production hold orders, date changes, excessive expediting, and failure of our priori- ty system) could also be corrected. Analysis and Objectives A review of our order cycle times showed that, although we planned on an average of six weeks, our cycle distribution centered around eight weeks. Further analysis of our labor his- tory indicated about one week's vouchered time spent on most orders, and seven weeks' worth of queue. In each of our 20 work centers, then, we averaged carrying between three and four days of orders waiting in queue. A portion of our seven-week queue time was required to balance production rates among operations; however, most of it existed for two questionable reasons: 1. It was the level of queue we had always maintained. 2. It was comfortable: Supervisors had rarely faced the problem of running out of work for their work center. Consider an alternative of half-day of queue per work center, which would allow for a total cycle of three weeks. is would allow us to ex- pect that the accuracy of customer scheduling, the overall responsiveness of the shop, and the severity of future order fluctuations should be improved. Methods With the help of our engineering staff and the production supervisors, a three-step plan was created to immediately stop releasing work to the shop for three weeks so that the new cycle time would be three weeks. 1. e immediate goal was to become cur- rent on delivery by authorizing overtime and shiing workers from those first work centers that now had to work. is was reached six weeks later; two weeks of our excess queue were eliminated. 2. Rather than face the disruption of releas- ing no work for three weeks, we decid-