APS — Advanced Planning and Scheduling
Advanced Planning and Scheduling (APS) is the class of planning software that schedules production against finite capacity. Classic MRP assumes every resource has unlimited capacity and every item a fixed lead time; APS drops both assumptions. It builds an executable schedule that respects machine and labor availability, sequencing rules, and material constraints — and it answers the question MRP cannot: not just what to produce, but in what order, on which resource, and when.
Where MRP stops
MRP nets demand against supply and proposes orders backward-scheduled by static lead times. That works as long as capacity is loose. Once a bottleneck work center is loaded past 100 percent, MRP keeps planning as if capacity were infinite, and the resulting plan is infeasible — due dates slip, expediting becomes the real planning system, and lead times get padded in self-defense. MRP II adds rough-cut capacity checks, but in most implementations it only flags overload and leaves resolution to the planner. APS takes the next step: it resolves the overload by moving, resequencing, or splitting operations until the schedule is feasible.
Finite-capacity scheduling and bottleneck resolution
An APS engine places each operation on a specific resource in a specific time slot, respecting shift calendars, planned maintenance, tooling, and operator qualifications. Most engines are bottleneck-oriented, drawing on Theory of Constraints logic: schedule the constraint resource first, protect it with buffers, and subordinate upstream and downstream operations to its rhythm. The practical output is a resource-level Gantt schedule the shop floor can actually execute, instead of a list of due dates that assume infinite machines.
Sequencing and setup optimization
Sequence-dependent setup is where APS earns its keep in many plants. Setup matrices encode that changeover time depends on the order of jobs: paint lines run light to dark, food plants schedule allergen-free products before allergen-containing ones, metal processors group jobs by alloy and gauge. APS builds campaigns that cut total changeover hours — and makes the trade-off against inventory and due-date performance explicit instead of leaving it to a scheduler’s whiteboard.
What-if simulation
Because an APS model contains resources, calendars, and routings, planners can simulate before they commit: What happens to existing promises if we accept this rush order? What does a weekend shift buy us? Which orders slip if machine 12 is down for two days? The same capability supports capable-to-promise quoting — a promise date based on finite capacity and material, one step beyond stock-based available-to-promise checks.
Embedded vs. add-on APS
US mid-market buyers face two sourcing routes. Embedded: many manufacturing ERP suites ship a finite-scheduling module on shared master data — no interface to build, one vendor, one user model, but often limited sequencing logic and optimization depth. Add-on best-of-breed: dedicated engines such as Siemens Opcenter APS (formerly Preactor), PlanetTogether, or Asprova offer deeper algorithms and richer scheduling boards, at the price of a permanent bidirectional interface — orders, routings, and calendars flow down; the resulting schedule flows back — usually complemented by shop-floor completion feedback from an MES. A useful rule: evaluate the embedded module against your three hardest scheduling scenarios first, and only go best-of-breed when it demonstrably fails them.
Data prerequisites
APS output quality tracks routing quality. The engine needs realistic setup and run times, complete setup matrices where sequence matters, accurate resource calendars, and timely completion feedback. Routings that were maintained for costing purposes rather than scheduling are rarely good enough; most APS projects include a routing cleanup phase before the schedule becomes trustworthy. Without closed-loop feedback, the schedule drifts from reality within days.
Selection criteria for US buyers
- Model your real constraints: run a pilot with your own routings, setup matrices, and a genuine bottleneck week. Generic demos hide the gap between a scheduling board and a scheduling solution.
- Planner experience: schedulers must be able to understand and manually override the schedule — drag-and-drop with immediate constraint feedback. A mathematically strong engine that planners bypass adds no value.
- Feedback loop: clarify how actual completions, scrap, and downtime reach the model — via MES integration, ERP confirmations, or manual entry — and how fast.
- Embedded first: price the embedded ERP module against the add-on including interface build and ongoing maintenance, not license against license.
- Rescheduling behavior: ask how the system handles disruption — full regeneration vs. repair-based rescheduling — and how stable the schedule remains for the shop floor when inputs change.
Comparable terms
MRP plans material quantities at infinite capacity; MRP II adds capacity checking; APS resolves capacity finitely. An MES executes and reports what APS schedules. Pull systems such as just-in-time and kanban shrink the scheduling problem in stable, repetitive environments — many plants run kanban loops for runners and APS for the variant-rich remainder. A scheduling model kept in sync with the physical plant is also a building block of a digital twin.