A scheduling application succeeds when the floor runs the schedule it produces. Nothing else about it matters.
The planner’s job is not to produce an optimal schedule. It is to produce one the plant will actually follow.
This table is why scheduling applications get bypassed. Every one of these constraints is real, the plant works around all of them every day, and none of them is in the system the engine reads from.
Capturing them is the project. The optimisation is the easy part once the constraints are honest.
Spreadsheet-driven production planning
The spreadsheet encodes every constraint in the table above, which is exactly why it has survived. Replacing it means capturing those rules first, not building a nicer interface.
Limited visibility into manufacturing capacity
Capacity is known as a theoretical rate rather than a demonstrated one. The gap between the two is where over-commitment comes from, and it is addressed in section 3.
Frequent production schedule disruptions
Disruption is normal. A scheduling application that assumes a stable week will be wrong by Tuesday, so recovery has to be a first-class function rather than a manual rebuild.
Material and resource availability constraints
A schedule built without checking material is a wish list. The material position has to be live, which makes this an integration requirement as much as a scheduling one.
Unplanned equipment downtime affecting schedules
When a machine stops, somebody has to decide what moves and what slips. That decision is the one worth supporting, and it needs to be made in minutes.
Inconsistent planning across plants and production lines
Each site plans its own way, often well, and a group application that forces one method will be resisted at the sites that already work.
Manual schedule adjustments and approvals
Adjustment is not the problem. Unrecorded adjustment is, because the reason for the override is the only information that would improve the rules.
Poor coordination between planning and execution
Planning publishes and execution reinterprets. Closing that gap requires the actual position flowing back, which is covered on the MES & Shop Floor Integration page.
Difficulty managing rush orders and priority changes
Every plant has an informal expediting process. It works and it is invisible, so its cost in changeovers and displaced orders is never counted.
Limited scenario planning and schedule visibility
Planners want to compare two options quickly. That is a different capability from offline capacity modelling, which belongs to Digital Twins & Simulation, and the distinction is in section 6.
Two thirds of the work in a scheduling project is establishing what is true: the constraints the plant actually operates under, and how long things actually take. Both are discovery exercises and neither is software.
Step 1
Production planning and scheduling process assessment
How the plan is made today, by whom, in what cycle, and what they consult. The planner’s own working file is the primary source document.
How the plan is made today, by whom, in what cycle, and what they consult. The planner’s own working file is the primary source document.
Step 2
Manufacturing resource and capacity mapping
Machines, tools, people and the skills each operation needs, including the shared resources that appear in nobody’s capacity model.
Machines, tools, people and the skills each operation needs, including the shared resources that appear in nobody’s capacity model.
Step 3
Planning constraints and business rules discovery
The table in section 2, filled in for your plant, with each constraint marked hard or soft. A constraint recorded as hard that is routinely broken will make the schedule infeasible.
The table in section 2, filled in for your plant, with each constraint marked hard or soft. A constraint recorded as hard that is routinely broken will make the schedule infeasible.
Step 4
Scheduling application architecture design
Where the engine runs, what it reads, how often it recalculates, and what the planner can change by hand.
Where the engine runs, what it reads, how often it recalculates, and what the planner can change by hand.
Step 5
Planning data and master data integration
Routings, work centres, calendars, bills of material and the material position, with a named owner for each. Master data quality sets the ceiling on schedule quality.
Routings, work centres, calendars, bills of material and the material position, with a named owner for each. Master data quality sets the ceiling on schedule quality.
Step 6
Scheduling rules and optimisation logic development
Built incrementally and validated against schedules the planner already considers good, so the engine earns trust before it is relied on.
Built incrementally and validated against schedules the planner already considers good, so the engine earns trust before it is relied on.
Step 7
Production planning workflow implementation
The publish, override, approve and recovery cycle, with the override reason captured as part of the workflow rather than as an optional comment.
The publish, override, approve and recovery cycle, with the override reason captured as part of the workflow rather than as an optional comment.
Step 8
ERP and MES integration
Orders and master data in, actual progress back, so the schedule reflects where production really is. The interface decisions are on the ERP and MES pages.
Orders and master data in, actual progress back, so the schedule reflects where production really is. The interface decisions are on the ERP and MES pages.
Step 9
Scenario testing and schedule validation
Run against historical weeks, including a bad one. A scheduler that only works in a normal week is of no use, because the schedule matters most when the week goes wrong.
Run against historical weeks, including a bad one. A scheduler that only works in a normal week is of no use, because the schedule matters most when the week goes wrong.
Step 10
Pilot deployment and continuous optimisation
One area, with the planner running both the old method and the new one until the new one is demonstrably better on their own terms.
One area, with the planner running both the old method and the new one until the new one is demonstrably better on their own terms.
Each is named with its precondition, because a scheduling capability deployed without its precondition produces output that gets overridden and then ignored.
Production planning applications
The planner’s working environment: demand in, capacity and constraints applied, a plan out. Requires an agreed planning cycle and a published freeze point.
Finite capacity scheduling
Scheduling against real capacity rather than infinite capacity. Requires measured times, per section 3. Without them it is precise and wrong.
Resource and machine scheduling
Machines, tools and people as separate constrained resources. Requires the shared-resource inventory, which usually exists only as a spreadsheet.
Material availability planning
Checking the material position before committing a sequence. Requires a live inventory feed and an agreed position on whether stock in transit counts.
Production order sequencing
Ordering within a period to reduce changeover and respect constraints. Requires the changeover matrix by product pair.
Constraint-based scheduling
Hard and soft constraints applied with a stated priority. Requires each constraint to be classified honestly, which is a planning decision rather than a configuration task.
Shift and workforce scheduling
Matching available people and skills to the plan. Requires a skills matrix and is subject to the employee relations considerations in section 7.
Schedule simulation and what-if planning
Comparing two or three options against real orders, quickly. This is a planning function. Offline modelling of a line’s capacity or layout belongs to Digital Twins & Simulation, and the two are distinguished in section 6.
Production schedule collaboration
Planning, production, maintenance and sales working on one published schedule with visible changes. Requires one published version, which is a governance decision.
Automated schedule alerts and exception handling
Notification when the plan and the actual diverge beyond tolerance. Requires actual progress flowing back from execution.
Multi-plant production planning
Allocating demand across sites. Requires agreed capacity definitions per site and a rule for who may commit another plant’s capacity.
The difficulty is named for each, because in scheduling the difficulty is almost always a constraint that is not written down rather than a calculation.
Daily and weekly production scheduling
The routine plan for the period ahead.
Machine and work-centre allocation
Deciding which machine runs which work.
Production order prioritisation
Deciding what runs first when not everything can.
Material shortage impact analysis
What becomes unmakeable when a delivery slips, and what to run instead.
Schedule adjustments following machine downtime
Rebuilding the plan when a machine stops.
Changeover and setup time optimisation
Sequencing to reduce total changeover.
Rush order scheduling
Inserting an urgent order and seeing what it displaces.
Workforce and shift capacity planning
Matching people and skills to the plan.
Multi-line production coordination
Sequencing across lines that share tools, people or materials.
Production schedule recovery planning
Getting back to the plan after a disruption.
A rule-based or heuristic scheduler that encodes the plant’s real constraints will beat a mathematical optimiser fed incomplete constraints, every time. Optimisation is worth adding when the rules are right and the remaining question is genuinely a trade-off between competing good options.
Production planning application architecture
The planner’s workspace, the engine, the published schedule and the floor’s view of it, with one authoritative published version.
ERP, MES and APS system integration
Orders, routings and material in; actual progress back. Where an APS product is already in place we integrate with it rather than replacing it by default.
Manufacturing resource and capacity data models
Machines, tools, people and skills as constrained resources, including the shared ones. This model is where multi-line feasibility is won or lost.
Constraint-based scheduling engines
Hard and soft constraints with stated priorities, and an explanation of why a given sequence was produced. An unexplainable schedule will not be trusted.
Operations research and optimisation algorithms
Applied where the problem is genuinely an optimisation and the constraints are complete. Introduced after the rules work rather than instead of them.
Heuristic and rule-based scheduling
The workhorse. Transparent, fast, adjustable by somebody who understands the plant, and sufficient for most plants most of the time.
Forecasting and demand data integration
Demand consumed as an input. The forecast itself is produced on the Predictive Analytics & Forecasting page and is not built here.
Scenario simulation and scheduling APIs
Comparing plans with real orders, and exposing the schedule so other applications can consume it rather than re-deriving it.
Cloud and on-premises application deployment
Planning can run centrally. The floor’s view of the published schedule should survive a lost link, which usually means a local cache at minimum.
Planning dashboards and workflow interfaces
The planner’s environment and the supervisor’s view are two different products, as they are across this group.
Governance defines who may change the schedule and the rules behind it, how exceptions get resolved, and how production continuity is protected. Everything below serves one of those three.
An engine the planner cannot override will be bypassed entirely. An engine they can override without recording why loses the only information that would have improved the rules. Both failures are common and the second is the more expensive, because it looks like success.
Manual override and planner approval controls
Overriding is a normal, expected action with a named authority, a reason captured from a short list, and a review of the reasons at a stated cadence. A reason code that is used constantly is a missing constraint.
Planning master data governance
Routings, calendars, work centres and the changeover matrix each with a named owner and a review cadence. Scheduling quality is capped by this data and it drifts quietly.
Scheduling rule ownership and approvals
Every rule has an owner in planning or operations, and a change to a rule goes through them rather than through a configuration ticket.
Production schedule version control
One published version at a time, with history, so a question about what the plan was on Tuesday has an answer.
Role-based planning access
Who may publish, who may override, who may only view. Publishing authority is the control that matters and it is usually too widely held.
Schedule change audit trails
What changed, when, by whom and why, retained so the pattern of change can be reviewed rather than only the current state.
Planning exception and escalation management
A defined route when the schedule becomes infeasible, reaching somebody who can authorise overtime, a deferral or a customer conversation.
Data quality and constraint validation
Automated checks before publication: a sequence that violates a hard constraint, a resource double-booked, a material commitment without stock. Better caught before the floor sees it.
Operational continuity and fallback procedures
A documented way to schedule when the application is unavailable, including who holds the current published schedule in a form that does not depend on it.
Planning model and algorithm governance
The rules and any optimisation logic documented, versioned and reviewed, with the ability to explain a given sequence. A scheduler nobody can explain cannot be defended to a supervisor.
Either one alone can be gamed. Adherence improves if the schedule is loose. Stability improves if nothing is ever re-planned. Reported together they describe whether the plan is both believed and followed.
Production schedule adherence
Did the plant run what was published, in sequence. The primary measure, and it has to be measured against the published version rather than the latest one.
Schedule stability and change frequency
How often a published period changed and how close to execution. Reported alongside adherence as the pair that cannot both be gamed.
Manufacturing capacity utilisation
Against demonstrated capacity rather than theoretical, per section 3. Against theoretical it is a number that always looks poor.
Resource and machine utilisation
Including tools and skilled labour, not only machines. The constrained resource is often not the machine.
Production order completion performance
On time and in full against the commitment, with lateness attributed to a cause: material, capacity, breakdown, or a change of priority.
Setup and changeover efficiency
Total changeover time per period and the share that sequencing could have avoided. This is where a scheduler most often pays for itself.
Material-related schedule disruptions
How often the plan failed for material reasons, which is usually a procurement or supplier finding rather than a planning one.
Schedule generation and recalculation time
It matters because of section 5: when a machine stops, a rebuild that takes twenty minutes will not be used by somebody on shift.
Planning accuracy and exception resolution
How often the plan was feasible as published, and how long exceptions took to close. Rising infeasibility points at master data drift.
Continuous scheduling optimisation
Override reasons reviewed and converted into rules, tracked as a count. This is the loop that makes the engine better, and it is the one most often left unclosed.
Improved production planning visibility, with one published schedule rather than several working versions
More realistic and achievable production schedules, which comes from measured times and captured constraints rather than from the engine
Reduced manual scheduling effort, once the rules are encoded. Before that, effort moves rather than falls
Better manufacturing resource utilisation, including the shared tools and skilled labour that are usually the real constraint
Faster response to production disruptions, which depends on recalculation being quick enough for somebody on shift to use
Improved coordination between planning and execution, which requires actual progress flowing back and is a joint outcome with your execution systems
If you already have a scheduling system that the floor reorders by hand, the fix is almost never the engine. Start with the constraint register and the routing time comparison.
CaliberFocus builds scheduling applications the floor actually runs. We capture the constraints that live in a planner’s head, measure real cycle and changeover times, start with transparent rules before adding optimisation, measure schedule stability alongside adherence, and make every override easy and recorded so the rules keep improving.
- A constraint register classified hard or soft
- Measured cycle times and a changeover matrix by product pair
- Rules first, optimisation where the trade-off is real
- A freeze window, with overrides recorded and reviewed
Security & Compliance
