Roles

Planning is the work package (what/how). Scheduling is who/when (usually the supervisor). The technician owns actuals and the observed failure mode. Reliability owns failure rate over time (RCM / FMEA / root-cause analysis). Fuse those jobs and the pager always wins.

NASA NPR 8831.2F already splits the functions in agency language: planners and estimators (P&E) prepare the work order (tasks, resources, safety, outages); scheduling then happens at three levels (shop load plan, typically in the work-control function; master schedule, usually under the senior shop supervisor; shop schedule, maintained by the shop supervisor who assigns work day-by-day). Breakdown / emergency trouble calls take precedence and are not a planner’s daily firefight. That is a US-agency procedure, not Palmer, but the split is the same shape.⁶

6.1 Planning versus scheduling (Palmer)

Doc Palmer (Richard D. Palmer, PE (Ret.), MBA, CMRP), Maintenance Planning and Scheduling Handbook (McGraw-Hill). AccendoReliability’s review by Mike Sondalini of the 3rd edition (labelled review, not the book): Palmer “separates planning and scheduling”; “for each, he gives 6 principles to work by” (principles not reproduced here — they live in the paid book). Planning is an entirely distinct step from scheduling. The planner sits out of the daily firefight so they cannot be pulled onto breakdowns. The planner is not the scheduler; the maintenance supervisor schedules. The planner produces a complete work package — purchased materials, tool list, procedures, drawings, past equipment history, job times, manning, external resources such as cranes — and walks away to the next work order. Breakdown jobs cannot be planned and go directly to the supervisor. Once a planned job starts, the crew solves in-job problems; the planner hears via the close-out and improves the next plan. Sondalini notes the book “starts quoting verifiable productivity improvements of 50% above the results achieved when not using planning” — that is the reviewer’s restatement of Palmer’s modelled case, not a multi-plant randomised trial.³¹

Prometheus Group (planner explainer, updated 1 Dec 2025 — vendor, labelled) uses the same what/how versus who/when split: the planner defines scope, resources and a step-by-step job plan; the scheduler assigns technicians and places the work on the calendar. Wrench-time losses they list: waiting on permits, hunting parts, walking to stores, isolation delays, bad job plans, rework, missing information. Their illustration: raising wrench time from 30% to 45% ≈ 1.5 extra technicians per five-person crew without hiring. Vendor illustration. They also cite “one planner for every 20 maintenance team members” as a “good rule of thumb” without Palmer’s assumptions — weaker than Palmer.³²

Palmer’s own modelled illustration, via ReliaMag quoting the handbook: if planning and scheduling raise wrench time from 35% to 55%, a 30-person crew could produce output equivalent to roughly 47. Modelled, not a multi-plant randomised controlled trial.³³

Planning exists to buy wrench time, not to write novels. SMRP defines wrench time as a work-management metric (hands-on / available time); the numeric “world-class” target lives in the paid Best Practices document and is not reprinted.²¹

6.2 Planner ratios — modelled, with assumptions; IDCON rejects one-size; SMRP 5.5.1 / 5.5.2 are definitions

ReliaMag (staffing page, fetched 28 Aug 2026) treats Palmer’s planner-to-technician ratio as “the best-documented rule of thumb in the field,” not a universally validated benchmark. Palmer is explicit about conditions:³³

Prometheus’s “one per 20” without those assumptions is the version that leaks into justification decks. Do not reprint it as Palmer.

IDCON (Torbjörn Idhammar): rejects one-size answers of the form “one planner per every 16 workers, or one supervisor every 20.” “It’s not that simple.” Count workload. Factors he names: whether the plant separates planners, schedulers and supervisors, shutdown versus daily planners, electrical/instrument versus mechanical planners; whether the planner is 100% planning or also buying, training workers on the CMMS, and sitting in improvement meetings; spare-parts / tool / technical-data / permit accessibility (if materials cannot be found, workload per work order rises and fewer jobs get planned); and crew independence (inexperienced crews consume more supervisor time). “The number of planners and foremen needed primarily depends on the level of expertise among the repair personnel, and how well the data and spare parts are organized.”³⁴

SMRP defines metrics 5.5.1 Craft Worker to Supervisor and 5.5.2 Craft Worker to Planner. ReliaMag: those are definitions (standardised calculation); the numeric ranges above are Palmer / IDCON, not reprinted SMRP targets. Do not write “SMRP says 1:20.”³³²⁰

Supervisor span: IDCON practitioner guidance ~8–10 where skills and work-management processes are weak, up to ~20 for independent crews with mature planning. ReliaMag: no universally accepted maintenance-specific target. McKinsey 2017 span-of-control archetypes (player/coach 3–5 through coordinator 15+) are general management, not maintenance, and McKinsey warns against a single universal span. Cite them only as that.³³³⁴

No universal technician-to-asset ratio. ReliaMag found none, for a structural reason: workload is driven by asset type, criticality, duty cycle, condition and strategy, not headcount of tags. NASA NPR 8831.2F staffs from a shop load plan that compares workforce composition to workload, using PM and PT&I resource estimates in work-hours by craft, and identifying personnel or skill shortages and excesses against that workload — a workload-based calculation, not a per-asset ratio.³³⁶

6.3 Technician owns actuals and the ISO 14224 failure mode

The EAM fails or works at close-out. The technician owns:

• Execute the package (or the emergency) safely. NASA: work proceeds to the approved work order; shops communicate with P&E when field conditions differ; a ~20% increase from estimate is NASA’s suggested threshold for a work-order amendment.⁶
• Record actuals at the point of work (mobile). IBM Think (vendor): hours, parts, notes in real time.² NASA 5.7.5.4: completion data on the form, supervisor review, then history files.⁶
• Select failure mode (observed effect) from a controlled list.

ISO 14224:2016 (catalogue abstract): comprehensive basis for collecting RM data in a standard format during the operational life cycle of equipment in petroleum, natural gas and petrochemical facilities; terms that constitute a “reliability language”; failure modes defined in the normative part as a “reliability thesaurus” for quantitative and qualitative applications; data-quality practices; minimum data covering equipment (taxonomy/attributes), failure (cause, consequence), and maintenance (action, resources, downtime). It does not cover direct cost data. Stated field of application is those industries; the taxonomy is widely borrowed. Wiki must not generalise the normative scope. Confirmed 2022; remains current.⁹

Consultancy explainer (iFluids, 20 Jun 2026 — not the standard text; use to navigate, then buy the PDF). ISO 14224 distinguishes:¹⁶

Conflating the three in one free-text box is, per that guide, the most common CMMS data-quality error. Mode is what RCM and failure-rate work usually run on. Forcing a controlled list at close-out is “the single most effective change a plant can make to data quality.” Four record types: equipment inventory, failure events, maintenance events, and population (count of identical units in service, with install/remove dates). Population is the denominator; omitting churn inflates MTBF. Nine-level taxonomy from industry down to maintainable item; each equipment class has a boundary diagram (do not code piping failures onto the pump). iFluids also cites a 2022 OREDA handbook update: centrifugal pump failure rates varied >40% between datasets with versus without strict subdivision discipline. Secondary; not reprinted as a fact until OREDA is opened.¹⁶

Maximo’s Problem / Cause / Remedy hierarchy is a product implementation of structured coding, not ISO 14224 itself. Map it; do not assume it is the standard.⁷⁹

Do not skip INPROG; do not close from memory at the workshop PC at shift end. That path destroys start clocks and mode quality. Chapter 8 collects the vendor write-ups (SAMEX, eWorkOrders, UpKeep) that describe the failure; the process rule belongs here.⁶⁷

6.4 Reliability owns rate over time — do not fuse with the pager

Fabrico (vendor blog, conceptually clean, labelled):³⁵

• Maintenance engineer / supervisor / technician: uptime now — repair quality, PM execution, backlog, MTTR.
• Reliability engineer: failure rate over time — FMEA / RCA, RCM, criticality, bad-actor elimination, design-out. Payoff is the failure that never happens.
• If one person holds both, the pager always wins. Ring-fence analysis time even if the headcount is shared.

Worked example in that article (vendor, logic is standard): third seal failure this quarter → maintenance swaps the seal in two hours (good MTTR); reliability finds misalignment at install across all three events and changes the procedure plus a laser-alignment check (the fourth failure does not happen). Measuring only MTTR rewards faster firefighting.³⁵

SAE JA1011 / RCM: reliability defines functions, functional failures, modes, effects, consequences, and task selection. The planner translates those tasks into CMMS job plans, intervals, and close-out fields (Diprose, Reliabilityweb). Loop: reliability → plans → structured close-out → better reliability. Free-text close-out breaks the loop. Diprose’s symptoms when the interface is weak: preventive tasks do not reflect actual failure modes; backlogs full of unvalidated work; technicians who see reliability tasks as low priority because the planning data does not make their purpose clear.¹⁴¹²

FMEA is the worksheet that often sits under RCM questions 3–5 (IEC 60812:2018 is the FMEA/FMECA standard — catalogue/webstore only here; do not invent worksheet steps from blogs).³⁶ RCA is after the event. Both starve without structured close-out.

MTBF vs mean time to failure (MTTF) (ReliaMag / reliability engineering): MTBF = repairable items; MTTF = non-repairable. Do not swap them. Availability ≈ MTBF / (MTBF + MTTR) under the usual identities. Exponential \(R(t)=e^{-\lambda t}\) assumes constant failure rate — not the bathtub’s infant-mortality or wear-out regions.²¹

GFMAM Landscape 6.3 Integrated Reliability is the Landscape home for reliability-through-life (failure modes, criticality, RCA, spares analysis). It is a subject of the discipline, not a definition of “APM software.”⁴

6.5 Who owns what in the EAM

Typical ownership (practice synthesised from Palmer / NASA P&E split, Diprose, Fabrico, SMRP metric names, ISO 14224 recorder roles). Not an ISO 55000 organogram.

Abdou Saied (LinkedIn, 2025-08, practitioner post — not a standard, not used as a citation of record): reliability builds the program (PM content, CBM system, models, design input); planner creates work orders and lines up execution. Useful only as a role-clarity check.

The failure mode of fusing reliability with the pager, and of using the planner as a clerk or firefight spare, is a program failure. Causes belong in chapter 8; the role split that prevents them belongs here.