No existing GMC matches this equipment well. Develop a new maintenance concept from scratch using the Concept Development tools on this page.
No results yet
Upload an O&M manual and/or describe the equipment, then click Match Concept.
Matched Concepts
Analysis Summary
No classification yet
Build the function tree above, then click Classify.
NORSOK Z-008:2024 §8
No FMECA yet
Fill in asset details and click Generate. Claude will propose failure modes per ISO 14224 for your equipment class.
ISO 14224:2016 Annex B · NORSOK Z-008:2017 §8.2.1
Generating FMECA — analysing equipment class against ISO 14224…
FMECA
ISO 14224:2016 Annex B · NORSOK Z-008:2017 §8.2.1
Criticality measure
Calibration
#
Subunit
Maintainable Item
Failure Mode
Code
Mechanism
Cause
Local Effect
System Effect
S
P
D
RPN
λ /10⁶h
Demands
Action
S = Severity · P = Probability · D = Detectability. All three are rated 1–10 against the anchor tables in the FMECA method guide; D runs in reverse (1 = certain detection, 10 = undetectable) · RPN = S×P×DARPN = S+P+DCriticality = consequence × likelihood; D not used · Failure mode codes are filtered per ISO 14224:2016 Annex B when the equipment type resolves to a Table A.4 class; where it does not, the full Table B.15 vocabulary is offered and nothing is filtered. Mechanism comes from Table B.2 and cause from Table B.3 — B.2.1 asks a failure record to carry all three, and B.2.3.1 warns they are easily conflated · λ and Demands from OREDA, enter manually · All cells editable.
On reading the RPN. The method is IEC 60812:2018 Annex B.4.2, which is informative — there is no such thing as an RPN-compliant analysis. S, P and D are ordinal ranks, so their product ranks failure modes but does not measure risk. The scale is not continuous: three 1–10 scales reach only 120 of the 1000 possible numbers, which is the standard’s own EXAMPLE 4 and is exactly what this tool produces. The ratio between two RPNs carries no meaning, and equal RPNs can describe very different situations. B.4.2’s own rule is that where RPNs are similar the higher-severity mode is addressed first — which is why S = 10 bands high on its own, marked nS. That rule is not cosmetic: it lifts 42 of the 100 maximum-severity combinations out of a band the bare product would have given them. Treat the column as a sorting aid and review S, P and D before acting on it. Switch to ARPN if the arithmetic bothers you — same three ranks, added instead of multiplied.
On reading the ARPN. The alternative risk priority number, IEC 60812:2018 Annex B.4.3 (informative). It exists because multiplying ordinal ranks is unsound: ARPN calibrates the three scales logarithmically — each level a fixed multiple of the one below, the same multiple for all three — and then adds them. Adding logs is multiplying the underlying quantities, so ARPN is the log of a relative risk, all 28 values from 3 to 30 are reachable, and one rank changing by one step always moves the result by exactly one. The standard’s worked example is S = O = D = 5 giving RPN 125 against ARPN 15; this tool returns those two numbers. Hover a value to see the risk ratio it implies. Two things to keep in mind: an ARPN only means something if your scale points really are a factor apart — set the calibration to match how you defined them and state which you used in the report, and note that over ten levels a ×10 multiple implies 10⁹ between the mildest and severest mode, which is rarely what a maintenance scale means. The severity rule still applies, so S = 10 bands high on its own. Where you hold real numbers for all three — OREDA failure rates, measured detection lead time — B.4.3 Note 2 says to calculate the risk from those directly rather than banding them at all. Switching method moves some rows. RPN and ARPN band identically on 854 of the 1000 combinations; the rest are the balanced middle cases RPN splits across a boundary despite identical ranks in a different order.
On reading the matrix. IEC 60812:2018 Annex B.3.2 (informative). Consequence against likelihood, each cell carrying a rank that is tied to a treatment — no arithmetic, so none of the ordinal problems RPN and ARPN have to work around. Detectability is not used. That is the standard's design, not an omission: B.1 gives D to RPN and ARPN only. For maintenance work it is usually the right default, because how easily a failure is caught is a property of the monitoring programme you are about to design rather than of the failure itself — folding it in now quietly rewards equipment you already watch and punishes equipment you do not. Expect rows to sit differently here than under RPN for exactly that reason — the two agree on 655 of the 1000 rating combinations, and the gap is D. The ten rating points are binned into five bands per axis (1–2, 3–4, 5–6, 7–8, 9–10), because a 10×10 grid is not a matrix anyone reads and B.3.2’s own examples use four to six categories. The rank names and the treatment each one triggers should be agreed before the analysis rather than after seeing the plot (B.3.1), and where a mode lands just either side of a boundary, check the two treatments really are different enough to justify the split.
No FMECA data
Complete Step 3 (FMECA) first to generate failure modes for the RCM decision tree.
Current failure mode
—
—
Consequence category:—override
—
RCM Strategy Summary
NORSOK Z-008:2017 §8 · ISO 14224:2016
#
Maintainable Item
Failure Mode
Code
RPN
Strategy
Justification
Complete Step 4 (RCM) first to define the maintenance strategy for each failure mode.
Generic Maintenance Concept
Draft — not submitted to Bluestream concept database
Bluestream · Generic Maintenance Concept · Draft
BS-GMC-??-??
Concept name
Equipment group · Type
Maintenance Plan
Maintainable Item
Failure Mode
Strategy
PM Type
Interval
Max allowed
Task Description
Comments
Basis & Rationale
Draft concept — not in the Bluestream library
This concept has been developed using the Bluestream Concept Development tool and is not persisted to the Bluestream GMC library. You can download the .docx for offline review, or use it directly to generate a Work Instruction. The concept is held in your browser for this session only and will not be saved to the library. To check if a matching library concept already exists, use the GMC Matcher tab on this page.
Work Instructions requires a completed GMC match. Run the GMC Matcher first, then return here.
Work Instruction Generator
Generates a structured, downloadable work instruction. Maintenance tasks come exclusively from the matched GMC Maintenance Plan lines. 1 token per document.
—
—
—
—
file.pdf
123 KB
Asset identity
Context
Fields missing. Without a Tag Number, the document will be generic rather than asset-specific.
Additional information needed
Clarification round 1
Refines the document to your context, no extra token cost.
Question from analysis
—
Ctrl + Enter to submit
Previous answers
Work instruction generated
Planner review required. Review the document before issuing to the field, verify hazard callouts, acceptance thresholds, and O&M references against current site procedures. The AI-generated footer in the document requests this explicitly.
No analysis yet
Describe the failure event, pick a method, then click Analyse Causes. Claude proposes a structured cause tree you can refine. Findings can be pushed back into your FMECA and RCM in one click.
IEC 62740 · Ishikawa 6-M
Analysing failure event — building cause tree…
Root Cause Analysis
IEC 62740 · Ishikawa 6-M
5-Whys causal chain
Fault tree
Build the tree from the top event down. AND means every input must occur; OR means any one is enough. Stop at a basic event when you would be guessing below it. Notation follows ordinary FTA practice.
The same basic event written under two branches is one event, and the analysis collapses it. That is how a shared power supply, a common instrument-air header or a single operator shows up as an order-1 cut set behind what looked like redundancy — the finding neither 5-Whys nor a fishbone can produce.
Fishbone — Ishikawa 6-M
Root causes
Immediate
Contributing
Systemic
Corrective & preventive actions
#
Type
Action
Owner
Due
Verification
Close the loop
Push findings back into the maintenance program so the same failure doesn't reach the same conclusion next time.
Drop a PM plan register to begin
The plan register is the only required file — one row per planned task, with a tag and an interval.
Completion history and failure history each unlock a further set of verdicts, and the tool tells you which
before you run it, not after.
Expect insufficient evidence to be the commonest verdict on a first upload. That is the honest
answer to a three-column export, and each one names the field that would change it.
Standard input is the D365 Asset fault cost control export with everything selected (Asset management → Inquiries → Asset fault → Asset fault cost control). Any spreadsheet with Asset, Fault, Criticality and a cost column will also work.
IEC 62740 · ISO 14224 · Operate · Tool 03
Bad Actor ranking
—
Cost per group Cumulative % 80% cut
#
Group
Events
Total cost
Share
Cum %
Top symptom · cause
Crit
Spare parts & holding policy
Category, location, max/min and re-order level — the four outputs
§12.2 asks for. Every figure below is recomputed as you type.
No spare parts yet. Add a line, or pull the consequence classes across from Step 01.
§12.1 makes that the starting point: the assessment shall be based on the
results of the consequence classification.
#
Part
OEM no.
Category
Consequence
Demand/yr
Leadwks
Unit costUSD
Location
Re-order
Order qty
Min / Max
Reference
NORSOK Z-008:2017 · Annex C
Generate Work Instruction
Pick the PM round to generate for. Longer intervals include all shorter ones.