Essential operations

Define the required function.

Describe the service the facility enables before listing assets. Agree its boundary, operating modes and measurable performance requirements. Normal operation, startup, shutdown, standby and degraded operation may create different needs.

• What must the operation deliver, to whom, and under what conditions?
• What minimum service is useful during disruption? What interruption, quality or environmental limits apply?
• Who can accept a degraded state, stop work or authorize restoration? Which decisions require specialist or regulatory review?

Record the source of each requirement and identify who must confirm it. A target proposed in a workshop is not an established design basis. Separate mission objectives from legal, safety and contractual obligations.

Adapted from TM 5-698-1's system requirements and RAM approach, and TM 5-698-4's system definition, ground rules and assumptions.

Dependencies

Trace the complete service path.

Connect the function to the systems, components, people and external services it depends on. A functional block diagram shows what each part does and how the parts interact. A reliability block diagram answers a different question: which paths must succeed for the required function to be delivered?

• Trace power, cooling, water, controls, communications, fuel, access, staffing, spares and supplier support as applicable.
• Identify single points of failure and shared dependencies. Two units can still share a switchboard, controller, fuel source, cooling system or maintenance error.
• Document transfer behavior, usable capacity, isolation and the effect of taking equipment out for maintenance. Do not assume a backup is independent or sufficient.

Use system drawings and operating knowledge together. Mark unknown connections explicitly. Quantitative reliability or availability models need appropriate failure data, repair assumptions and treatment of common-cause failures; simple diagram arithmetic is not proof of performance.

Method basis: TM 5-698-1, chapters 2–4; TM 5-698-4, chapter 3. Keelstead adds organizational and supplier interfaces to the planning record.

Evidence

Establish what is known.

Collect the information needed to test the requirements and model. Useful sources include current drawings, equipment data, inspection and test reports, maintenance history, failure logs, operating procedures, training records and actual disruption experience.

• Record each source's date, revision, equipment or system scope and responsible reviewer.
• Distinguish an observation, a reported value, an assumption and a calculation. Keep conflicting evidence visible.
• Check whether test conditions represent the operating mode being assessed. A successful component test may not demonstrate an integrated service path or extended outage capability.

Record missing data as an open item. When historical data are sparse, use a documented qualitative review rather than inventing precision. If generic data are used, explain their applicability, uncertainty and limitations.

Method basis: TM 5-698-1's reliability data and analysis principles; TM 5-698-2, chapter 2 and section 5-3. Evidence IDs and review-status fields are Keelstead adaptations.

Critical gaps

Analyze failure and choose a response.

Failure Modes, Effects and Criticality Analysis (FMECA) examines how a function or item can fail, the resulting effects and their importance. Work from the agreed boundary and operating mode, not a generic list of bad outcomes.

• State the failure mode, credible cause and local, next-level and mission effects.
• Identify detection methods and existing compensating provisions. Record what must work for a mitigation to be effective.
• Assess consequence and likelihood using a consistent, documented method. Quantitative criticality requires defensible data; qualitative judgments need definitions and rationale.

Review severe consequences, hidden failures, common-cause exposure and uncertainty explicitly. A risk-priority number can help organize a discussion, but must not bury a serious consequence or imply that a threshold authorizes safe operation.

Reliability-Centered Maintenance (RCM) then asks which tasks can preserve the required function or manage a specific failure mode. Consider condition-based work, scheduled restoration or replacement, failure-finding tasks and justified corrective maintenance. A task needs technical applicability and a worthwhile consequence or cost benefit. If maintenance cannot adequately address the risk, consider design, operational or other changes.

Do not assign intervals from a generic template. Qualified personnel should establish the task, method, frequency, competency and acceptance criteria using the failure behavior, applicable requirements and equipment guidance.

FMECA basis: TM 5-698-4, chapters 3–6. RCM basis: TM 5-698-2, chapters 4–5. This summary and the Keelstead records do not replace the full analyses.

Action plan

Implement, verify and learn.

Translate the analysis into a manageable set of decisions and work packages. Separate immediate evidence gathering, maintenance changes, operational preparations and capital options. Define what would close the gap before marking an action complete.

• Assign an accountable owner, due date, required reviewer and decision authority.
• Record dependencies, resources, cost range, residual risk and the evidence needed for acceptance.
• Use controlled change and qualified review for equipment, procedures or maintenance changes. Plan verification without introducing uncontrolled operational risk.

Review whether the completed action changed the original failure scenario. Update drawings, procedures, training and the underlying analysis as needed. Feed incidents, near misses, maintenance findings and exercises back into the record.

Method basis: TM 5-698-2's life-cycle RCM program; TM 5-698-4's recommendations; UFS 3-520-02's prepare, absorb, recover and adapt framing. The integrated action register is Keelstead's adaptation.