12.3 Creating Production and Service Schedules

Once loading philosophy and PAC methods are clear, schedulers must create the actual sequence of work. In production, that means ranking jobs at a resource. In services, that means assigning appointments, crews, or tickets to limited slots. CPIM treats both as detailed scheduling problems: finite capacity, competing due dates, and rules that trade one performance measure for another.

Priority Rules You Must Know

Common dispatching / priority rules include:

• FCFS (First Come, First Served) — Process jobs in arrival order. Fair and simple; weak when a long job blocks many short ones.
• SPT (Shortest Processing Time) — Run the shortest job next. Tends to minimize average flow time and average number of jobs in the system; may make long jobs very late.
• EDD (Earliest Due Date) — Run the job with the nearest due date next. Tends to reduce maximum lateness and improve on-time performance; ignores processing length.
• CR (Critical Ratio) — Priority index = (due date − today’s date) / remaining processing time (in the same time units). CR < 1 means the job is behind; lower CR is more urgent. CR updates as time passes and work completes.

Other rules (least slack, fewest operations remaining) appear in practice, but FCFS, SPT, EDD, and CR are the Domain VII core.

Worked Comparison: Four Jobs at One Work Center

Assume today is Day 0. Four jobs are ready at a single machine. Processing times and due dates (in days) are:

FCFS sequence: A → B → C → D
Completion times: A=4, B=6, C=11, D=12
Lateness (completion − due, positive = late): A=0, B=0, C=1, D=4
Average flow time (completions, all ready at 0): (4+6+11+12)/4 = 8.25 days
Jobs late: C and D (2 late)

SPT sequence: D(1) → B(2) → A(4) → C(5)
Completions: D=1, B=3, A=7, C=12
Lateness: D=0, B=0, A=0, C=2
Average flow time: (1+3+7+12)/4 = 5.75 days
Jobs late: C only (1 late)

EDD sequence: B(due 6) → D(due 8) → C(due 10) → A(due 12)
Completions: B=2, D=3, C=8, A=12
Lateness: all ≤ 0 (none late)
Average flow time: (2+3+8+12)/4 = 6.25 days
Jobs late: 0

Critical ratio at Day 0 (CR = due / processing, since today = 0 and remaining time = full processing):

Lowest CR first: C (2.00), then A and B tied at 3.00—break ties with a secondary rule (here EDD): B before A, then D. Sequence: C → B → A → D
Completions: C=5, B=7, A=11, D=12
Lateness: C=0, B=1, A=0, D=4
Average flow time: (5+7+11+12)/4 = 8.75 days
Jobs late: B and D (2 late)

What the numbers teach

SPT won on average flow time. EDD won on due-date performance in this set. FCFS was “fair” but mediocre. Static CR at a single moment did not dominate—and CR must be recalculated as days pass. If the machine runs C first and we reach Day 5 with B, A, D still waiting, new CRs change priorities. Exam questions often ask which rule minimizes average flow time (SPT) or which best protects due dates in a simple set (often EDD).

Takt Time: Pacing Production to Demand

Takt time is the available production time divided by customer demand rate for a period:

Takt time = Available working time / Demand quantity

Example: A shift offers 450 productive minutes and demand is 150 units → takt = 450/150 = 3 minutes per unit. The line should complete a unit about every 3 minutes to match demand without overproducing. Cycle time of the bottleneck operation must be ≤ takt; if a station’s cycle is 3.5 minutes, the schedule cannot sustain demand without overtime, parallel resources, or demand leveling.

Takt is a design and pacing tool, not a priority rule. It shapes how many stations, how much WIP between stations, and whether the daily schedule is feasible. Mixed-model lines use takt with heijunka (leveling) so the sequence of models still averages to takt.

Service Appointment Scheduling Parallels

Service operations—clinics, field repair, consulting desks—face the same finite-capacity logic:

• Slots are like machine hours; overbooking is infinite loading by another name
• Arrival order versus shortest expected duration versus earliest promised time mirrors FCFS, SPT, and EDD
• No-shows and variable service time create queue growth similar to scrap and run-time variance
• Takt thinking appears as appointments per hour matched to demand (for example, 6 intake slots per hour if average service is 10 minutes and two clinicians work)

A field-service dispatcher who always sends the nearest technician (travel minimization) may ignore the earliest SLA due time—exactly the SPT-versus-EDD tradeoff. A clinic that books every open minute without recovery slots will see waiting rooms explode when one complex visit overruns—failure of finite loading and I/O-style control of arrivals.

Building the Schedule in Practice

Creating the schedule combines:

1. Demand or order pool for the horizon
2. Capacity calendar (machines, clinicians, crews)
3. Priority policy aligned to order winners
4. Setup or changeover logic (or appointment type grouping)
5. Takt or appointment-density check for feasibility
6. Publication of a dispatch list or appointment book, with freeze rules for the near horizon

Domain VII does not require you to run a full APS engine. It requires you to interpret small numeric examples, pick the rule that matches the stated goal, compute takt, and recognize that services use the same capacity math with different vocabulary.