Scheduling
The Finishing Department manager was out sick so Milliken decided to tour there the next day. Back in the Production office Milliken sat down with Jim Worthen, the plant scheduler, and asked him about his job.
Worthen told Milliken that his job was to do whatever it took to deliver on time. The plant held a daily production meeting every morning where the main topic was what had gone wrong the previous day, the status of large, important orders, and what the department managers would do to deliver on time any newly expedited orders. Worthen indicated that generally they did not tell Sales if an order was going to be late. He said, "We're supposed to but a lot of times we don't know for sure until the end because we're hoping to get a partial out on time. If you tell Sales ahead of time an order is going to be late, Wascov just starts hammering us and calling up Dave Rand and the rep starts screaming. It creates a lot of unnecessary work—and then they bump something else, so the ripple effect is just more jobs late. Of course, they don't blame themselves for that, do they?"
Milliken asked, "When you partial, do you break orders in half or have a little part and a bigger part usually? And do you break into more than two pieces?" Worthen said, "It's always just the first and second piece, and they're just about always the same size—it's easier that way to justify doing all the additional setups if the partials don't have tiny numbers."
Milliken said, "Sean Quinn and Rick Gomes say they can't use the schedule. Why is that?" Worthen told him that much of the shop floor reporting was missing or wrong, forcing Worthen to spend several hours each day trying to scrub the data using information on work order pieces completed in each operation which the operators recorded by hand in the Work Order Jackets.
Milliken asked why the data reported through the shop floor computer terminals was so inaccurate. Worthen told him that operators recorded pieces completed in the Work Order Jacket and their start and stop times on setup and run. The Estimator used that data for estimating standard times, and Payroll got attendance data when the operators swiped their bar-coded ID cards through the terminals. Milliken knew that operators got a 30-minute meal break in the middle of their 8-hour shift. Roll stock use at the sheeter, plates in Comp, and ink in Print were recorded by different, manual systems, and the experienced operators there maintained their own local inventory records to make sure that they did not run short. As a result, no one felt a real need for accurate terminal-based shop floor production reporting.
Milliken asked what rules the computer-based scheduling system used. Worthen told him that the computer recalculated and printed a new schedule twice a week for two "buckets"—the machine capacity in hours available in each work center in the Monday-Tuesday-Wednesday and the Thursday-Friday-Saturday (half-day) buckets. The system added up the standard setup and run hours for every order released to the plant and "filled the buckets" by scheduling orders by priority.
Priority was set by the "critical ratio". The ratio's numerator was the number of hours between the calculation date and the due date, and the denominator was the total standard setup and run hours for all remaining operations plus 48 hours for each remaining operation. The lower the ratio, the higher the priority. Worthen added, "My rule of thumb is that if the critical ratio when the job starts is lower than 2, I have to watch that job, it's born in trouble."
While Worthen was explaining this, Milliken sketched his understanding of the rules with an example (see below). He showed this to Worthen, who nodded and remarked, "Nobody ever got that before the first time I told him."