Work orders, PM, and inventory

The work order is the transactional object that makes enterprise asset management (EAM) software a system of record rather than a logbook. NASA’s facilities-maintenance procedure defines it as the document directing shops to perform maintenance work, carrying task requirements (usually by craft), labour, material and equipment estimates, coordinating instructions, and administrative and financial information.¹ IBM Think (vendor article, labelled) treats work orders as “the system of record for all maintenance and service activity,” with the data contributing to asset history, performance and lifecycle decisions.² Those two statements are the whole operational game: the same record authorises spend, isolation, kitting and assignment, and later is the history (cost, duration, failure). ISO 55000:2024 does not define work-order fields, PM generation rules, or storeroom min/max; those sit in maintenance-execution practice and in the product.³

The Global Forum on Maintenance and Asset Management (GFMAM) Landscape v3 places this work in subject 6.5 Maintenance Delivery: identification, planning, scheduling, execution and analysis of maintenance work through the intended life cycle, with strategies that “align to the asset life cycle stage, criticality, and risk.”⁴ SMRP’s Guide to the Maintenance and Reliability Body of Knowledge names Pillar 5: Work Management as the professional-body home for the same cluster.⁵ Neither document is a software specification.

5.1 Control document and history record

NASA NPR 8831.2F Chapter 5 is a complete public work-control specification (US agency procedure, not ISO). Work is received, classified, identified, estimated, approved, scheduled, tracked, accounted for, analysed and reported from inception to completion; the computerized maintenance management system (CMMS) is the control system for that path. Each item of work is given a unique identifier that “should not be changed once it is assigned.” Status at a minimum covers work reception, planning and estimating, final authorisation, scheduling, material management, work performance and final inspection — not all milestones apply to every job (trouble calls skip most of them).⁶

IBM Think (vendor, labelled) describes a six-step process: submit a request; approve, prioritise and create the work order; plan, schedule and assign; execute and track; complete and document; review and analyse. A standard template in that article includes requester, asset ID and location, priority, description, and required parts, tools and safety procedures. Mobile actuals (hours, parts, notes) are how the record becomes usable later.²

If the work order is not linked to an asset or location, cost and failure do not roll up. Maximo-Users.net’s 2026 work-order practice note (practitioner / Maximo-community blog, labelled) puts this as the first rule: a work order without asset or location is an orphaned record — actuals will not roll up to asset cost history, failure reports will not populate, and reliability analysis has blind spots. Treat that as Maximo-community practice, not a standard. NASA Appendix E makes the same hierarchy point in agency language: work orders can be written against a location or against equipment at a location; “using locations allows for tracking the equipment’s life cycles (history).”⁷⁸

Commissioning and retirement must hit that same register, or later mean-time-between-failures (MTBF) arithmetic is fiction: ISO 14224:2016’s population (in-service count with install/remove dates) is the denominator. That argument is in chapter 3; it is restated here only because the work order is how failure events and maintenance events attach to the population.⁹

5.2 Types are plant-defined (keep 5–7); there is no ISO list

There is no single ISO list of work-order types. Plants invent codes. EN 13306:2017 (Maintenance — Maintenance terminology, CEN/TC 319) is the European vocabulary for maintenance types and activities (preventive including predetermined and condition-based; corrective after fault recognition, at catalogue-family level). It is a terminology standard, not a software taxonomy, and software-only maintenance is outside its scope. Catalogue via BSI / SIS; text paywalled. Do not treat EN 13306 as a work-order-type pick-list.¹⁰¹¹

Two independent public taxonomies show the point:

NASA NPR 8831.2F classifies work by work element (agency usage): preventive maintenance (PM) and PM finds; predictive testing and inspection (PT&I) and PT&I finds; grounds care; programmed maintenance (PGM); repair; trouble call (TC); replacement of obsolete items (ROI); service request; central utility plant operations and maintenance. Those categories “extend beyond the minimum required for financial accounting” and exist so managers can see where resources go. NASA also classifies by funds type, approval level, special interest, size and method of accomplishment. That is one organisation’s schema, documented in a free US-government procedure.⁶

Practitioner consensus around Maximo-shaped plants (Maximo-Users.net 2026, labelled community) converges on a short operational list of 5–7 types, not twenty: define the taxonomy in a governance document; enforce it in workflow or UI; “limit work types to 5–7 categories — too many creates classification fatigue and inconsistent data.” Without consistent types, PM compliance and emergency-versus-routine spend cannot be calculated.⁷

A typical short list, as industry usage, not a standard:

IBM Think (vendor) names an overlapping but different set as “common types”: corrective, reactive, preventive, predictive, installations, inspections, audits, logistics. That is a product-class illustration, not a taxonomy to copy.²

Enforcement, not slogans. Type must be mandatory at creation. Service-level-agreement clocks belong on reactive types, not on PM. Material-costing rules differ if one work order covers many assets (bulk inspection) versus one asset (parts must hit one cost object). MaximoMastery (March 2026 explainer; practitioner blog, Cloudflare-gated on direct fetch — cited from a search extract of that URL only): the common mistake is using corrective (CM) for everything, which erases planned versus reactive; do not inflate PM counts; track emergency separately.[7a] Maximo-Users.net’s backlog dashboard lists an “emergency-ratio target <10%”; that figure is not sourced to SMRP in that article — do not reprint the 10% as a standard.⁷

5.3 Status is not decoration

NASA requires the CMMS to hold estimated or actual start and completion dates and the responsible party for each applicable milestone. Work status is “a key element in maintaining good customer relations”; skipping milestones is not a reporting convenience.⁶

A typical Maximo-shaped status flow, from Maximo-Users.net 2026 (product-specific community practice, widely copied; IBM’s official Work Order module page was not opened this pass — do not quote this table as IBM-official):⁷

Skipping INPROG (APPR → CLOSE) loses start/end times. Maximo-Users.net: that skip “loses visibility into work actually in progress and the real actual start/end times needed for MTTR calculations.”⁷ NASA 5.7.5.4 requires completion reporting into the CMMS — resources used, unanticipated conditions, a concise description of work accomplished, additional material not listed on the order — with shop-supervisor review.⁶ Status codes exist so those clocks and costs are real.

IBM Maximo product behaviour (not a process principle): PM records generate work orders via Generate Work Orders or a cron task (`PMWoGenCronTask`). IBM support’s troubleshooting note for PM work-order generation names the cron-task class `psdi.app.pm.PMWoGenCronTask` and the PM record’s eligibility checks (active PM, operating asset/location, seasonal dates, job-plan status). Cancelling a PM-generated work order can offer to reinstate the PM. Cite as product behaviour; the Maximo Manage Preventive Maintenance HTML was 403 this pass and the support note is labelled IBM support, not a process principle.¹³[13a]

5.4 What must be on the record (actuals in the first 90 days)

NASA 5.6.1: once requested work is approved, the work order is the document directing execution. It includes an estimate of resources (work hours by craft, materials, equipment, tools, specialised support); the steps or tasks; coordination and outages; safety requirements (confined space, lockout/tagout, oxygen depletion, chemical or explosive handling, fall protection, training/certification); job priority; and accounting information.⁶ Small jobs (NASA: typically less than about 20 work hours) may skip detailed planning, but the craft supervisor still specifies safety requirements.⁶

Practitioner consensus (Maximo-Users.net 2026 work-order note + enterprise CMMS note, both community, labelled), aligned with IBM Think’s template and with Reliabilityweb (Diprose) on close-out:²⁷¹⁴¹⁵

• Asset or location (always).
• Work type and priority, from a criticality × impact matrix, not “whoever shouted.” NASA Table 5-2 is a sample six-level priority system (emergency / urgent / priority / routine / discretionary / deferred), not a global standard.⁶
• Job plan (product language: IBM Maximo “job plan”; other products: procedure, standard job) for repetitive work: tasks, craft/hours, materials, personal protective equipment / lockout/tagout, drawings. Maximo-Users.net claims organisations using job plans report 30–40% less planning time for recurring work. Community claim; no method shown. Do not quote as a measured industry result.⁷
• Materials reserved or issued against the work order (see §5.7).
• Actuals before close: labour hours, materials, start/end. Maximo-Users.net enterprise CMMS guide (community, labelled): “Organizations that fail to establish actuals discipline within the first 90 days of go-live rarely recover it — the habit either forms early or does not form at all.” Without actuals, cost reports, job-plan accuracy and asset life-cycle cost are fiction.¹⁵
• Failure coding on corrective work. Problem / cause / remedy (Maximo product hierarchy) or mode / mechanism / cause (ISO 14224) — structured, not free text. See chapter 6. Maximo’s triplet is a product implementation of the same idea, not ISO 14224 itself.⁹¹⁶
• Long description for the job; short description as headline. Maximo-Users.net’s “40–60% fewer calls to planning” from long descriptions is another unsourced community figure — do not reprint as a result.⁷

NASA 5.7.5.4 already requires the technician-side completion data on the form, with supervisor review. That is the public, non-vendor statement of the same discipline.⁶

5.5 Preventive maintenance is a strategy, not a cron

IBM Maximo Manage product documentation (Preventive Maintenance topic; product; HTML 403 this pass — restated from a search-index extract of that URL, last updated 2026-01-09 on that extract, plus IBM Think’s preventive type): PM is a strategy for scheduling recurring work to reduce unplanned downtime. PM records are templates that generate work orders on elapsed time, meter units, or both; job plans supply the task content; generation can be restricted to days/months/seasons; a PM hierarchy is supported. That is execution. Reliability strategy (what should be on the PM) lives outside the cron task.¹³²

EN 13306 family (catalogue-level; clause text paywalled): preventive work is done to assess or mitigate degradation and reduce probability of failure. It includes predetermined (interval, no condition investigation) and condition-based. Corrective is after fault recognition. Software-only maintenance is out of scope.¹⁰¹¹

RCM decides what belongs on the PM. Nowlan and Heap (Reliability-Centered Maintenance, United Airlines work published by the US Department of Defense, 1978) established that most failure modes are not age-related in a way that makes blanket overhaul effective, and that some PM induces failure. A PM that does not address a failure mode is waste or risk.¹²¹⁷

SAE JA1011 AUG 2009, Evaluation Criteria for Reliability-Centered Maintenance (RCM) Processes (SAE International; 1999 original, revised August 2009), does not prescribe a software tool. It states the minimum a process must meet to be called RCM. Conscious Reliability’s practitioner restatement (labelled secondary — buy SAE for the legal text) gives seven steps in order: operating context, functions and user-desired performance standards (verb + object + quantitative standard, not nameplate); functional failures (total or partial); failure modes reasonably likely to cause each functional failure; failure effects; failure consequences (hidden vs evident; safety/environmental vs operational vs non-operational economic — pick the most severe); proactive tasks and intervals that are technically feasible and worth doing; default / one-time changes (redesign, procedure, training, operating context) when no task is feasible and worth it. Task families in that restatement: on-condition / condition-based maintenance (CBM) (P-F interval must fit); scheduled overhaul; scheduled discard; failure-finding (hidden functions, protective devices); redesign. Interval math must be acceptable to the owner; RCM cannot override law or contract.¹²¹⁸

RCM is not “add more PMs.” It often deletes time-based tasks that have no failure-mode justification. Vendor blogs that quote “RCM eliminates 30–40% of time-based tasks” did not show a named study on the pages fetched for this draft — do not reprint.

How a PM program actually works in an EAM (process, not a product wizard):

1. Strategy (reliability): which failure modes, which JA1011 task type.
2. PM record in the EAM: asset/location, frequency or meter, job plan, seasonal windows, hierarchy.
3. Generation on a rolling horizon (not the morning it is due). Maximo-Users.net (community) recommends a 4–6 week horizon so planners can kit and schedule; nightly generation for PMs due within N days; weekly PM-backlog review.⁷
4. Plan / kit / schedule like any other work order.
5. Execute to a checklist with acceptance criteria, not “check pump.”
6. Close with findings. Findings spawn CM. The yield metric lives here.
7. Review: if nothing ever fails between intervals, extend or kill the PM; if failures cluster, shorten or change tactic.

Maximo-Users.net enterprise CMMS guide (community, labelled): original-equipment-manufacturer (OEM) intervals are conservative (worst-case, zero failure-cost tolerance). Failure history should drive frequency. Vague tasks (“inspect motor”) produce inconsistent results; put numbers on the task (“vibration at bearing housing 0–4 mm/s RMS; escalate >6”). Meter-based PMs need a meter-reading discipline first. They state “80% monthly PM compliance is the industry minimum.” That 80% is vendor/community, not the SMRP 5.4.14 best-in-class (above 90%). ReliaMag rates vendor “below 80% = broken” and “95%+ on A-class” as low-confidence vendor conventions (e.g. eWorkOrders).¹⁵¹⁹

Do not inflate PM. If a technician noticed lubrication was needed and did it without a schedule, that is corrective work, not PM. Relabelling reactive work as PM poisons planned-work percentage and hides the firefight. An oversized PM library crowds out real work and destroys compliance. Right-size from failure history and RCM / failure modes and effects analysis (FMEA), not from “more PMs = more mature.” Reliabilityweb (Chris Diprose, Planner HQ): when planning is disconnected from reliability, PM tasks do not reflect actual failure modes; technicians treat reliability tasks as low priority because the plan never says why.¹⁴

5.6 Work-management key performance indicators (KPIs) — formulas yes; targets with provenance

SMRP Best Practices (7th edition is the member download observed 28 Aug 2026) standardises 70+ metrics with definition, formula, sample calculation, best-in-class target values, and cautions. The PDF is member-gated / copyrighted. Do not paste unpublished numeric targets from memory. SMRP organises metrics under five pillars, including work management. ReliaMag (independent explainer) maps the usual set onto that pillar and onto EN 15341:2019+A1:2022, and refuses unsourced “world-class” numbers.²⁰²¹

EN 15341:2019+A1:2022 (Maintenance — Maintenance Key Performance Indicators, CEN/TC 319; BSI published the British adoption 31 Mar 2023 / Intertek Inform 29 Mar 2023) lists KPIs of the maintenance function and gives guidelines to select a suitable set, to appraise and improve effectiveness, efficiency and sustainability of existing physical assets (industrial, infrastructure, facilities, civil buildings, transportation, etc.) in the framework of external and internal influencing factors. ReliaMag: eight groups — one for physical asset management, six for maintenance sub-functions (health, safety and environment; maintenance management; people competence; maintenance engineering; organisation and support; administration and supply), and one for information and communication technologies. It superseded EN 15341:2007 and was amended in 2022. Complements SMRP; do not assume identical formulas. Catalogue via BSI / Intertek Inform; text paywalled. Do not dump indicator codes from sample PDFs.²²²³²¹

Working formulas (standard engineering expressions as ReliaMag states them; the governing standard wins for audit):²¹

SMRP forum (Michael Raible, Best Practices Committee chair): Guideline 5.0 defines work types; relevant metric IDs include 5.4.1 Reactive, 5.4.2 Proactive, 5.4.12 PM & PdM Yield, 5.1.13 PM & PdM Effectiveness, 5.3.1 Planned Work, 5.4.8 Planned Backlog, 5.4.9 Ready Backlog. The committee has not published quartile benchmarks — only best-in-class. Do not invent quartiles.²⁰

Palmer vs SMRP on schedule compliance — two named positions; do not mix with PM compliance. ReliaMag (PM compliance page): SMRP sets a high target for schedule compliance; Doc Palmer argues a week that routinely clears ~90% was under-loaded, because a normal plant absorbs ~20% urgent break-in, so ~80% on a fully loaded schedule is realistic.¹⁹

Palmer in Plant Services (primary, named): schedule compliance is a management metric, not a supervisor score. His worked definition uses sheer number of work orders: if a schedule gives 100 orders, the crew completes 60 of those plus 140 others, compliance is 60/100 = 60% (not 60/200 and not 200/100). A score below ~40% usually means the supervisor ignored the schedule; a score above 90% usually means management did not give enough work (Parkinson’s Law). “Considering that even a pretty good plant still has about 20% reactive work, it’s not really practical to mandate that supervisors achieve schedule compliance above 80%.” He cites John Crossan (SMRP Annual Conference 1997): “schedule compliance is the ultimate measure of proactive maintenance.” In a follow-up: the purpose of scheduling is not high compliance in itself (schedule one job, get 100%); it is to complete more proactive work than would otherwise die in the backlog. Fully load next week’s labour capacity and “take our lumps.” If you routinely achieve 90% or higher on a supposedly full load, “we are simply deluding ourselves.” It is okay for the first-line supervisor to break the schedule for genuine reactive work; it is not okay for management to ignore what the score is pointing at.²⁴²⁵

Gaming. Auto-close overdue PMs, back-date completions, or loosen the on-time window and “compliance” rises while MTBF falls. Pair 5.4.14 with yield and a reliability outcome. ReliaMag: a 95% rate sitting next to a falling MTBF deserves a look; chasing 100% can signal an under-sized PM program or a window so loose that “on time” means little.¹⁹

ISO 22400 (manufacturing operations KPIs, including overall equipment effectiveness (OEE)-related indicators) sits alongside SMRP and EN 15341; a metric labelled the same way in a vendor dashboard and in EN 15341 may be calculated differently. When the number will be audited, use the source standard.²¹

5.7 Inventory — the bill of materials must be on the work order

Maintenance, repair and operations (MRO) inventory is not production inventory. Demand is intermittent and lumpy; standard economic-order-quantity-on-forecast often fails. The EAM’s job is to tie parts to assets and to work orders, then replenish from that demand — not to be a second enterprise resource planning (ERP) system.

NASA NPR 8831.2F 5.7.2: material management includes ordering, stocking, storing, staging, issuing and receiving material for use on work orders. Planners and estimators prepare material-requirements lists as part of the work order; the material manager obtains the material and advises schedulers when it is available. Range and depth of stock rest on historical demand, standby items (spares for critical systems), and projected requirements; high/low limits from use rates, economic reorder quantities and delivery times. Chapter 6 of the same NPR lists materials/tools files among typical CMMS functions: inventory, assign materials to work orders, requisitions.⁶²⁶

GFMAM Landscape subject 6.8 Supply Chain Management is the delivery-side provisioning loop (purchasing, inventory, warehousing, contracting) aligned with asset-management objectives, taking total cost of ownership into consideration. Subject 3.7 (Resourcing Strategy and Management) covers people, tools, materials and services needed to deliver asset-management plans. Do not collapse 6.8 with a CMMS storeroom screen.⁴

Practitioner guides (CPCON spare-parts guide, 27 Mar 2026; Verdantis storeroom guide; Forthcast MRO guide — all vendor/consultancy, labelled) agree on the data model even where their return-on-investment numbers are vendor-side:²⁷²⁸²⁹

• Item master: one stock-keeping unit per unique part. Duplicates (“PART”, “MISC”, same bearing five ways) fragment stock and wreck reorder math. Maximo-Users.net (community): dedupe before turning inventory on.¹⁵
• Bill of materials (BOM) / spare list on the asset: every maintainable asset lists its parts and approved substitutes. Without this, planners cannot kit and technicians cannot see on-hand from the asset card.
• Work order → materials: planned parts on the job plan; reservations for scheduled jobs; issues charged to the work order; receipts against purchase order. Bypass purchasing = invisible cost (Maximo-Users.net tip 11).⁷
• Reorder: min/max or reorder point (ROP) from actual lead time + demand variability, not catalogue quotes. CPCON (vendor): quoted versus actual lead times often diverge; using quoted lead time under-protects. Usable as process advice; not a measured constant.²⁷

Classify before stocking. Do not ABC-only. A $5 gasket can stop a line.

Intersect them (VED × ABC × XYZ). Vital + cheap + intermittent still stocks one. Desirable + expensive + intermittent often does not. Part categories in common industry usage (CPCON, labelled): critical spares, insurance/capital spares, consumables, rotables (repair-and-return), kits. Rotables need a separate installed-versus-spare tracking loop.²⁷

Policies that match the class (practice, not ISO):

• Vital, long lead time: stock based on risk, not forecast. Shared pools / vendor-hold where sites share the failure mode.
• Fast consumables: min/max, two-bin, vendor-managed inventory / consignment.
• Slow non-critical with short lead time: buy on the work order, do not sit on the shelf.
• Slow-moving and obsolete (SLOB): Verdantis (vendor) rule of thumb — review items with no movement 24–36 months unless tied to critical/legacy equipment. CPCON (vendor): equipment retirement must trigger BOM review or obsolete stock accumulates. Labelled vendor rules of thumb, not SMRP.²⁸²⁷

Formulas (standard inventory theory, restated by CPCON and Forthcast; not proprietary):²⁷²⁹

• ROP ≈ demand during lead time + safety stock.
• Safety stock when demand and lead time vary: \\(SS = Z \\sqrt{LT \\cdot Var(D) + D^{2} \\cdot Var(LT)}\\). Differentiate Z by VED; do not apply a 99% service level to rags.
• Economic order quantity is for stable consumables, not Z-class insurance spares.
• Intermittent demand: Croston / Syntetos–Boylan — named methods; do not pretend a 12-month moving average works on a four-year turbine part.

Storeroom discipline (Verdantis, vendor): every issue against a valid work order; receipts against purchase order; planners reserve for the schedule. Cycle count by ABC. System quantity ≠ shelf quantity means reorder points are theatre.²⁸

Inventory KPIs — treat published numeric targets as vendor unless SMRP is opened. ReliaMag lists “stocked MRO inventory value as % of replacement asset value (RAV)” as an SMRP business-and-management metric and again refuses to reprint the paid target.²¹ CPCON’s 27 March 2026 spare-parts guide (vendor/consultancy, labelled) publishes a table of stockout, fill-rate, MRO-turns, carrying-cost, obsolescence and emergency-purchase-order targets, and cites Plant Engineering’s 2025 maintenance survey and Aberdeen MRO-as-%-of-RAV figures in the prose. Those survey numbers were not independently verified here and are not reprinted. They are not SMRP. ReliaMag’s refusal to reprint the paid SMRP MRO-%-RAV target is the wiki rule: name the metric, do not invent or second-hand a percentage.²⁷

TRM (IBM/TRM executive panel write-up, 18 Mar 2026, Will Goetz — vendor-reported process claim): without clean hierarchies, BOMs, criticality and inventory data, asset performance management / AI cannot deliver. That claim is process, not a number — usable if attributed. The panel’s plant-outcome anecdote is vendor-reported and is not used here.³⁰