Production Planning Excel Migration | APS Guide for Thai Factories
A production planning Excel migration is not a campaign to ban spreadsheets. It is a redesign of where the approved plan, constraints, exceptions, approvals, and plan-versus-actual differences are governed. This guide gives Thai factories a practical sequence for moving toward APS while retaining Excel for analysis: migration signals, system boundaries, minimum data, a 90-day pilot framework, RFP questions, and acceptance tests.
The first misconception in a production planning Excel migration
Excel remains a capable analysis and simulation tool. Microsoft 365 supports simultaneous co-authoring of compatible workbooks stored in OneDrive or SharePoint Online, and Version History can be used to review or restore earlier versions. The claim that Excel cannot support collaboration is therefore inaccurate.
Collaboration, however, is not the same as production-plan governance. Show Changes can track edits to cell values and formulas, but Microsoft notes that some changes are not shown, including actions involving charts, shapes, PivotTables, formatting, hidden cells, and filters. Gaps may also arise when older Excel versions, unsupported features, file replacement, or copied files are involved.
A factory needs to govern more than the identity of the person who edited a cell. It needs to know:
- which demand, inventory, capacity, and calendar version supported the plan;
- which plan version is approved and when its frozen period begins;
- which due-date, material, capacity, setup, or alternate-resource constraint was violated;
- who approved an exception for a rush order or equipment outage, and why;
- how deviations between plan and actual are returned to the next planning cycle.
Leaving Excel therefore means managing planning elements as business objects, not merely changing a file extension. Excel may remain useful for hypotheses, personal analysis, or meeting support. What must not remain ambiguous is whether the approved master plan resides in Excel or APS and where the shop floor receives authoritative instructions.
Microsoft lists worksheet limits of 1,048,576 rows and 16,384 columns, a cell-content limit of 32,767 characters, and 200 adjustable cells in Solver. A 32-bit Data Model shares 2 GB of virtual address space with Excel and its add-ins. Although 64-bit Excel has no fixed file-size ceiling, practical use still depends on memory and system resources. These numbers should not be treated as automatic reasons to abandon Excel. Even a small workbook becomes risky when versions, exceptions, approval, constraint propagation, and actual feedback are uncontrolled.
For a broader diagnosis, see when Excel-based manufacturing management reaches its limits.
Signals that justify moving to a production planning system
Look first at operating symptoms, not file size. If several of the following conditions have become routine, it is reasonable to evaluate a production planning system or APS pilot.
| Signal | What happens in daily work | What the migration must design |
|---|---|---|
| Versions proliferate | Files named final, final2, and planner copies coexist | System of record, version, approval state, release time |
| Changes are frequent | Every rush order, shortage, or outage triggers broad manual edits | Events, impact range, replanning rules |
| Frozen periods move | Near-term confirmed work changes without explicit approval | Time fences, permissions, exception approval |
| Bottlenecks are hidden | Local schedules look feasible while work accumulates elsewhere | Finite capacity, queues, movement, WIP |
| Alternate resources depend on one expert | Only a veteran knows which machine or tool may substitute | Alternatives, priority, quality qualification |
| Actuals do not return | Yesterday’s delay or yield difference is absent from today’s plan | Actual capture, variance, replanning trigger |
| Adjustments cannot be explained | Sequence changes are justified only as planner intuition | Reason codes, rules, audit trail |
Do not frame these symptoms as a failure of the Excel planner. The planner may have used formulas, colors, notes, and secondary sheets to represent constraints that formal systems never captured. The workbook is not just a disposal target; it is a source of requirements. Interview the reasons behind added columns, color rules, morning adjustment sequences, and confirmation calls, not only the formulas.
Separate the roles of MRP, APS, ERP or production management, and MES
A common automation error is to treat material requirements and finite-capacity scheduling as one calculation. Separate them by decision responsibility rather than product label.
| Domain | Primary question | Typical inputs | Main outputs |
|---|---|---|---|
| ERP or production management | What should be made or purchased, in what quantity, and by when? | Orders, demand, inventory, BOM, procurement rules | Requirements, orders, inventory and cost records |
| MRP | Which materials and components are needed, how many, and when? | Demand, BOM, inventory, lead time | Net requirements, proposals, requirement dates |
| APS | Under constraints, which resource should perform each operation and in what sequence? | Orders, routing, capacity, setup, calendars | Finite schedule, sequence, load, delay and conflict |
| MES and shop-floor actuals | What actually started, completed, stopped, or became defective? | Dispatch instructions and resource events | Start and finish, quantity, downtime, quality, timestamps |
SAP’s official PP/DS process distinguishes infinite-capacity planning in MRP from subsequent finite-capacity sequencing and resource planning with heuristics or optimization. It considers requirements, receipts, inventory, resource availability, and component availability, and supports interactive resolution of delays and overloads. This is a useful illustration of role separation, not evidence that SAP is the only answer.
Siemens describes Opcenter APS as supporting strategic, tactical, and detailed scheduling, including resource allocation, order priority, batching, capable-to-promise decisions, bottlenecks, and setups. Vendor functionality is not a universal business-result guarantee. A factory still needs to validate its own data quality, decision cadence, and operating responsibility.
ANSI/ISA-95.00.01-2025 addresses the boundary between enterprise and logistics functions and manufacturing-control domains, the information shared across that boundary, and the value of standardized integration interfaces and semantic models. It does not mandate a specific product. It provides a useful frame for defining the responsibilities of ERP, APS, production management, MES, controls, and operations.
For each data object, specify where it is created, approved, distributed, and updated with actuals. Avoid dual masters, such as maintaining resource calendars independently in ERP and APS or allowing the shop floor to overwrite APS instructions in an unofficial workbook.
Minimum data set before moving to APS production planning
APS does not automatically repair the constraints it receives. Define each data object and its owner before the pilot. The entire factory does not need perfect data on day one, but missing values must be visible, provisional values must have an approver, and production use must have a quality gate.
| Data | Minimum validation | Example owner |
|---|---|---|
| Items | Unit, status, validity, make or buy | Production control and item governance |
| BOM | Quantity, yield, substitutes, effective dates | Engineering and process engineering |
| Routing | Sequence, standard time, overlap, subcontracting | Process engineering |
| Resources and capacity | Machines, people, tooling, fixtures, concurrent load | Manufacturing and maintenance |
| Calendars | Shifts, breaks, holidays, maintenance, overtime | Manufacturing, HR, maintenance |
| Setups | Changeover conditions, cleaning, color, tool changes | Manufacturing and quality |
| Alternate resources | Eligibility, preference, quality qualification | Manufacturing and quality |
| Inventory and WIP | Availability time, hold status, lot, location | Warehouse and production control |
| Orders and demand | Due date, priority, split permission, confidence | Sales and production control |
| Actuals | Start, finish, quantity, scrap, stop reason | Manufacturing and MES owner |
An owner is not merely a person who types data. Ownership covers definition, approval, update frequency, quality threshold, and exception decisions. In a Thai operation, headquarters may approve item and BOM data while the local plant maintains shifts and capacity. That cross-site division should be explicit.
Break data quality into completeness, validity, consistency, timeliness, duplication, and traceability. Standard times may be populated for every item yet still reflect an obsolete machine condition. An alternate machine may be present in the master but unusable because its quality qualification has expired.
When reviewing existing workbooks, inventory values, formulas, named ranges, validation, colors, comments, hidden columns, macros, and external links. Do not recreate everything automatically. Classify each element as required by law, contract, or quality; required for a planning decision; or merely a display preference.
Validate production planning automation with a 90-day pilot
This article proposes 90 days as a practical pilot framework, not a guaranteed implementation duration. Adjust the period for the plant, process, data, and integration scope. Start with a bottleneck process or one product family where planning difficulty and business relevance can be observed, rather than every item at every plant.
Phase 1: Freeze the current plan and evaluation conditions
Preserve the legacy plan used for comparison. Capture the input timestamp and versions of orders, inventory, WIP, calendars, shortages, and priorities. Record the sequence of decisions used by the Excel planner.
Turn the pilot into observable questions:
- Can the same plan be reproduced from the same inputs?
- Can assignments beyond bottleneck capacity be detected?
- Can approved frozen periods be protected?
- Can the impact of a shortage or outage be explained?
- Can the reason and approver for a manual change be traced?
- Can actual variance be returned to the next cycle?
Phase 2: Build the constraint model
Load items, BOM, routing, resources, capacity, calendars, setups, and alternatives only for the selected scope. Decompose spreadsheet logic into constraints, priority rules, data transformations, and presentation instead of copying formulas mechanically.
Distinguish hard constraints that must not be broken from soft constraints that may be relaxed with authority. An unqualified machine may be prohibited, while a setup-minimization rule may be relaxed for a genuinely urgent order. Without this distinction, the model may produce no plan or a plan the factory refuses to execute.
Phase 3: Run Excel and APS in parallel
Do not switch off the existing workbook immediately. Generate both plans from the same input snapshot. When results differ, classify the cause instead of declaring a winner.
| Difference | Question | Response |
|---|---|---|
| Data | Are inventory, capacity, time, and units identical? | Correct extraction, conversion, and cutoff time |
| Rules | Are priority, setup, and freeze policies interpreted the same way? | Agree on constraints and operating rules |
| Tacit judgment | Is this an exception known only to the planner? | Document the rationale and approval process |
| Presentation | Is the same result shown at different granularity? | Standardize comparison views and aggregation |
| Not executable | Are people, machines, or tools unavailable in reality? | Correct the resource model and actual feedback |
Do not assume the spreadsheet is the truth. It may contain stale values or tacit corrections. Compare both plans with shop-floor evidence and judge which decisions are reproducible and explainable.
Phase 4: Test exceptions deliberately
A schedule for normal orders is not sufficient acceptance evidence. Inject a rush order, material shortage, equipment outage, setup change, overtime cancellation, and prohibition of an alternate resource. Verify both the system response and who is authorized to decide the exception.
Phase 5: Decide acceptance and the next stage
Use evidence agreed before the demonstration. Classify open points as configuration, data, interface, training, operating rule, or product limitation. Then decide whether to expand, run another pilot, revise requirements, or stop.
For deeper scenario design, see how to run production planning simulations.
RFP questions and a complete cost breakdown
An RFP should enable vendors to be compared using real data and exceptions, not a feature checklist. Terms such as AI optimization or automatic scheduling are insufficient. Ask about inputs, constraints, outputs, reproducibility, explanation, and operational ownership.
Core vendor questions
- Where do MRP and finite-capacity planning separate, and what data passes between them?
- Can machines, labor, tools, fixtures, and components be modeled as simultaneous constraints?
- How are setup matrices, batches, campaign production, and alternate resources configured?
- Can changes inside a frozen period be blocked or routed for approval?
- Can the same input, configuration, and software version reproduce the same plan?
- Are the reason, user, and time of post-schedule manual changes traceable?
- How far can the impact of shortages, outages, and rush orders be explained?
- What are the ERP, MES, and equipment interface methods, frequency, retry, and error handling?
- What Thai, English, and Japanese interface, master data, training, and support are available?
- Who changes the model, and how are testing, production release, and rollback governed?
Compare more than license fees.
| Cost area | Scope to confirm | Common omission |
|---|---|---|
| Licenses | Users, sites, resources, functions, environments | Test systems, viewers, extra modules |
| Implementation | Requirements, model, screens, reports | Exception rules and revision rounds |
| Data preparation | Extract, transform, cleanse, initial load | Shop-floor confirmation, provisional approvals |
| Integration | ERP, MES, equipment, files, APIs | Monitoring, retry, time and unit conversion |
| Testing | Unit, integration, performance, acceptance, exception | Test data, evidence, retesting |
| Training | Planners, operators, administrators, support | Multilingual content, shifts, new hires |
| Support | Incidents, updates, backup, inquiries | Local hours, holidays, change governance |
| Change management | Roles, meetings, KPIs, procedures, adoption | Conditions for retiring the Excel master |
This article does not state a market price because scope, data quality, interface count, constraint complexity, and number of sites change the estimate. Compare assumptions, exclusions, customer tasks, and triggers for extra cost alongside price.
Use the production scheduler comparison for Thai manufacturing to extend the product-selection criteria.
Acceptance testing for a production planning system
An application opening and a schedule button running are not sufficient. Prove the full flow from input and processing through result, explanation, access control, release, and actual feedback.
Normal-flow tests
- Import approved demand, inventory, BOM, routing, and calendars correctly.
- Apply expected cutoff time, time zone, units, and rounding.
- Produce a plan that considers capacity, components, setups, and alternatives.
- On rerun, reproduce the result or explain the difference.
- Prevent unauthorized edits to an approved plan.
- Release the plan to operations and return actuals to the next cycle.
Exception-flow tests
| Exception | Expected observation | Acceptance evidence |
|---|---|---|
| Rush order | Impact on existing due dates, materials, and capacity is visible | Before and after plan, warning, approval history |
| Material shortage | Unavailable period and affected orders are visible | Shortage list, delay reason, substitution decision |
| Equipment outage | No assignment during downtime and replanning is possible | Resource calendar and replan log |
| Setup change | Sequence-dependent time and prohibited combinations apply | Setup setting, sequence, warning |
| Overtime cancellation | Capacity reduction propagates to load and due dates | Calendar version, load variance, delays |
| Alternate resource prohibited | Unqualified equipment is not assigned | Constraint, violation detection, approval evidence |
Keep the input snapshot, configuration version, run time, output, expected result, approver, and unresolved issues. A screenshot alone cannot establish reproducibility or distinguish a defect from a data difference.
Define performance by business scenario rather than a universal number: daily scheduling, same-day replanning, capable-to-promise response, and interactive meeting simulations may need different response times. Measure not only computation time but also the time needed to detect an exception, understand its reason, secure approval, and publish the revised plan.
Operating design for an APS rollout in Thailand
A translated interface does not create multilingual operations. Establish a controlled dictionary for item names, operations, downtime reasons, setup conditions, and approval reasons. Decide which of Thai, English, or Japanese is authoritative. Combine coded reasons with optional explanation so variants in free text do not prevent analysis.
Define storage time, display time, and interface time separately. Headquarters in Japan and a Thai factory may interpret the same date differently because of cutoffs and night shifts. Include local holidays, company holidays, shifts, planning-day boundaries, and connections to regions with daylight-saving time in the tests.
Document responsibilities such as:
- who confirms demand priority between headquarters and the Thai plant;
- who maintains local capacity, holidays, and maintenance stops;
- who approves urgent changes inside the frozen period;
- who is accountable for releasing the APS plan to operations;
- who records night-shift outages and shortages and triggers replanning;
- who monitors interface failures among ERP, APS, and MES.
Thailand’s BOI reported 132 applications worth approximately THB 17.2 billion in the first half of 2026 under the Smart and Sustainable Industry measures, covering machinery upgrades, digital technology, and automation or robotics integration. These are nationwide application figures. They do not demonstrate the benefit, approval, or tax incentive of an individual APS project.
Current BOI guidance states a minimum efficiency-improvement investment of THB 1 million, excluding land and working capital. General conditions describe import-duty exemption on machinery and, for existing businesses, a three-year corporate income tax exemption capped at 50% of the efficiency-improvement investment. The cap conditions may differ where qualifying machinery is linked to Thailand’s domestic automation industry. Eligibility depends on the activity, investment content, rules at the application date, and BOI’s assessment. Confirm the individual case with BOI or a qualified adviser before applying.
Failure patterns and countermeasures for leaving Excel planning
Making spreadsheet removal the goal
People may continue using Excel for supporting analysis or rapid checks. A ban tends to create hidden spreadsheets. Define the system of record, permitted auxiliary use, whether re-import is allowed, and the retirement condition.
Waiting until every master record is perfect
Perfection delays learning. Limit the scope, expose missing and provisional values, assign owners and correction dates, and prevent provisional data from crossing the production gate without approval.
Evaluating only the optimization engine
A strong calculation is not an operating process if actuals do not return, exception reasons are missing, or the result cannot be released. Test the end-to-end flow.
Starting with all plants and all items
An excessive scope mixes data, model, and governance problems. Learn from a bottleneck or product family, define rollout conditions, and then expand.
Treating vendor improvement rates as a company guarantee
Vendor outcomes depend on customer conditions. Establish your own baseline, input, period, exclusions, and measurement method. This article does not guarantee an improvement rate for delivery, inventory, or productivity.
Trying to eliminate human adjustments
Exceptions require judgment. The goal is to govern authority, reason, impact, approval, and reproducibility while preserving useful shop-floor knowledge and protecting the master plan.
Frequently asked questions
Does a production planning Excel migration mean banning Excel completely?
No. Excel may remain for analysis, hypotheses, and supporting tables. Define the authoritative location for approved plans, constraints, exceptions, and actual variance, and prevent an unofficial workbook from overwriting it.
What is the difference between a production planning system and APS?
Product terminology varies. ERP or production management generally covers orders, inventory, BOM, requirements, and work orders. APS emphasizes finite-capacity assignment and sequencing under resource and setup constraints. Verify inputs, decisions, outputs, and ownership rather than labels.
Does APS make MRP unnecessary?
Not necessarily. MRP calculates component requirements; APS addresses finite capacity and sequence. Design how they exchange requirement dates, capacity information, and plan versions.
Should a detailed scheduling pilot include every item?
Usually not at first. Select a bottleneck or product family with meaningful constraints, due-date impact, shop-floor support, and a comparable legacy plan.
What are acceptance criteria for production planning automation?
Core criteria include reproducibility with the same input, constraint-violation detection, frozen-period protection, traceable change reasons, and return of actual variance to the next cycle. Add plant-specific exceptions.
Is migration unnecessary if the workbook is below Excel’s row limit?
No. Technical limits are context, while versions, frequent change, exceptions, approval, actual integration, and dependence on individual knowledge drive the governance decision.
Does a 90-day pilot complete production deployment?
Ninety days is a suggested framework, not a promise. Data, interfaces, constraints, approvals, and multi-site scope may shorten or extend it. The pilot should produce evidence for the next investment decision.
Does an APS investment automatically qualify for BOI incentives?
No. Eligibility varies by activity, investment, current rules, and BOI assessment. Confirm the individual application with BOI or a qualified adviser before submission.
Conclusion
Moving production planning beyond Excel is a governance and operating-design project, not a file-removal project. Define the system of record, constraints, exceptions, approvals, and actual variance; then connect the responsibilities of MRP, APS, ERP or production management, MES, and operations. Use existing workbooks as requirement evidence, pilot a bottleneck or product family, retain the old plan for parallel comparison, and test difficult exceptions. Expansion should follow evidence of reproducibility, constraint detection, frozen-period control, reason tracking, and actual feedback, not an unsupported promise of numerical improvement.
If your Thai factory is still mapping Excel planning, preparing an APS RFP, designing a 90-day pilot, or clarifying ERP and MES interfaces, talk with TOMAS TECH. We can start by visualizing your current system of record and responsibility boundaries.
References
- Microsoft: Excel specifications and limits
- Microsoft: Collaborate on Excel workbooks with co-authoring
- Microsoft: Show Changes in Excel
- ISA: 2025 update to ISA-95
- SAP Help: Production Planning and Detailed Scheduling process
- Siemens: Opcenter Advanced Planning and Scheduling
- Thailand BOI: 2026 first-half Smart and Sustainable Industry applications
- Thailand BOI: Smart and Sustainable Industry
