Machine scheduling with a Gantt chart

Use it when more than one machine is involved: several machines doing the same work (injection molders, presses, looms) where you must decide which one takes each job, or a job shop where each job visits different machines in different orders (saw, lathe, mill, drill). For one machine or the bottleneck, the job sequencing calculator is simpler and shows the family setup rules in detail.

Standard time = job quantity × time per part (time study or routing) ÷ actual efficiency. Full setup is the measured changeover from one family to another; the short one is the adjustment when the family is already set up. Relative speed comes from each machine’s actual hourly output on the same product. Allowed machines come from the mold, tooling or part size.

There are 9 jobs on 3 injection molders, one of them 20% slower and free only at 9:00. The order entered finishes at 22.50 h with 3.50 h of tardiness on one job. EDD (earliest due date first) delivers every job on time and finishes at 26.00 h. LPT finishes first (21.50 h) but makes 4 jobs late: it balances the load, it does not protect due dates.

There are 6 jobs going through saw, lathe, mill and drill. The mill is the bottleneck with 14.50 h of load: no schedule can finish earlier. MWKR reaches exactly 14.50 h (the optimum) but makes two jobs late; EDD finishes at 16.00 h with only 1.00 h of tardiness on one job. The wait table shows where each part queued and behind which job.

It is the common-sense shop floor rule: a free machine with a part waiting starts work right away. When several parts are queued, the chosen rule decides which goes in (SPT, EDD, CR, MWKR or the order entered). It is the simplified Giffler–Thompson method used in scheduling systems; it gives good schedules but does not guarantee the optimum.

Not necessarily. Scheduling several machines is a hard problem: the number of combinations grows very fast and the exact optimum needs mathematical programming. The rules on this page give a good arrangement in seconds that is easy to explain. The lower bound shows how much could still be gained: if the rule reaches it, the schedule is optimal for makespan.

All jobs are released at the start, with no breakdowns, no material shortages and no splitting a job across machines. With parallel machines, setup depends only on the previous family; in routings, each operation’s setup is fixed. The calendar is the same for every machine, without weekends or holidays, and there is no operator limit. For hundreds of jobs with labor or tooling constraints, use a scheduling (APS) system.