15-Session Programme · 25 Minutes Per Day · Free · For the owners and operators of Nigeria’s energy assets
WEEK 2 OF 3 · INTELLIGENCE IN THE FIELD AND THE FIRST LIFECYCLE CHOICES · SESSIONS 6–10 · WORK AT YOUR OWN PACE
Chapter 6 · Intelligence in the FieldSESSION 6
Reliability Begins with the Failure Mode
Why This Matters
In Session 5 you made inspection answer a decision. Today the question is whether a warning can do the same. Predictive maintenance is an attractive phrase because it suggests that failure will announce itself early enough for the organisation to respond. Sometimes there is a useful detectable pattern. Sometimes the failure is too sudden, the measurements too weak or the action window too short. You need to know which situation applies before you approve a prediction programme. By the end you will be able to define the event, name the simpler method a model must beat, read a validation result and test whether a warning can become a feasible action in your company.
25-Minute Module
0:00 – 5:00
Start with the functional failure
- Equipment can carry a defect and still do its job. A functional failure is the point at which it can no longer perform a required function at its specified standard. Then name the mechanism or mode the application is meant to address. A vibration model aimed at one mechanical condition should not be credited with preventing stops caused by power, feed conditions or an unrelated protective trip.
- Define what counts as a failure, an alert (the system's flag that something may be developing) and a successful intervention (the response the team agrees to count as having worked). Distinguish loss of required function from a minor defect, a planned stop and a nuisance alarm.
- Record operating exposure and duty. A machine that runs intermittently should not be compared with one in continuous duty using calendar counts alone.
- THE DOCTRINE
Before approving a prediction programme, establish whether the failure gives a useful detectable pattern, or is too sudden, too weakly measured or too quick for the action window.
Ask which failure, exactly, the model is for. Then ask what else stops the same machine.
5:00 – 10:00
Honest records and a credible simpler method
- Data can be biased by successful maintenance: if a team intervenes before failure, the record no longer shows what would have happened. Recognise that limit rather than inventing labels to complete a training dataset. A work order saying repaired does not establish the mechanism; record what was found, what was changed and whether the investigation supported the diagnosis, or the model learns administrative vocabulary instead of equipment behaviour.
- Compare the model with the method it would actually improve: an existing condition limit, an engineering rule, a specialist review or a better-maintained instrument. That is the Sequencing Discipline from Session 1. Comparing AI with no action makes the benefit look larger than the real investment choice allows.
- THE DOCTRINE
Do not let one aggregate accuracy percentage carry the argument. Evaluate detected events, missed events, false alerts and the lead time available for an actual response, using definitions the team can reproduce.
- In a dataset dominated by normal operation, a system that predicts no failures can look accurate while giving little useful warning. False alerts consume scarce specialist time and damage trust; a very quiet model may miss important events.
A model that beats doing nothing has not yet beaten anything you would actually do.
10:00 – 16:00
A fictional validation exercise
- The book's teaching case, with invented inputs that are not a reported field result. A retrospective test contains 120 defined review windows: 12 with the target event and 108 without it. The model flags nine of the event windows and 18 normal windows. It misses three events and correctly leaves 90 normal windows unflagged.
Measure Question it answers Calculation Result Flagged windows containing the event When it flags, how often is the event there? 9 ÷ 27 (9 + 18 flagged) 33.3% Event windows detected Of the events, how many did it catch? 9 ÷ 12 75% Normal-window false-positive fraction How often does a normal window raise a flag? 18 ÷ 108 16.7% Overall classification accuracy How many windows were classified correctly? 99 ÷ 120 (9 + 90) 82.5% - The figures answer different questions. They do not establish a field false-alarm rate per machine-month, because the number and duration of windows would have to be defined for that. They do not establish economic value or acceptable safety performance. Ask whether the nine detections arrived early enough, whether the 18 false alerts generated unnecessary work and why the three events were missed.
- If most detections were already obvious to the existing condition-monitoring team, the incremental benefit may be small. A lower headline accuracy could still be useful if it warns of a costly condition the present method misses. The operating context decides, not the largest percentage.
I would ask which question each percentage answers before letting any of them into a board paper.
16:00 – 21:00
Validate in time, then make the warning actionable
- Windows next to each other in time look alike, so randomly splitting overlapping time-series windows can leak similar information into both development and evaluation sets and flatter performance. Ask the technical team how it prevented that.
- Test the regimes that matter: startup, shutdown, low load, changed feed and modifications. A model trained before an overhaul may no longer represent the equipment. The evaluator should be able to reproduce the test with the identified data and version.
- Lead time is useful only in relation to the response. A two-day warning may be valuable when the team can inspect that shift, and insufficient when the replacement has a long manufacturing and mobilisation cycle.
- AT THE ASSET IN THE FIELD · THE WORKED CASE
- In the field. A fictional gas plant's reliability engineer receives an early warning on a compressor. The alert playbook, written with operations, sets the initial checks, escalation authority, review deadline, permitted interim action and recording. The protective systems carry on exactly as before.
IN THE BOARDROOM · THE PARALLEL- In the boardroom. The same company's executive committee is asked to fund a wider prediction programme. The useful question is the response path, not the headline accuracy. Where response capacity is limited, improve triage and readiness before adding alerts. Same loop, seen from the budget.
- THE DOCTRINE
Keep protective systems and established emergency procedures independent of an advisory model. The advisory may be unavailable or wrong at the moment it is most needed.
Ask what the team would physically do in the first shift after the alert.
21:00 – 25:00
The charge, and your wrap
- THE LEADERSHIP CHARGE. Require the target failure mode, the simpler comparator and the response window before approving predictive maintenance. Evaluate the entire decision pathway.
- Sometimes the honest answer is that the failure mode gives no useful warning, or that the constraint sits in execution, not prediction. Then a prediction programme is the wrong investment.
- Before you close: write the target failure mode for your Session 1 decision, the method your company uses today and the response window. That is Section 6 of your 90-Day Intelligent Asset Plan. Write your assignment now.
A well-founded decision not to buy a prediction programme is good ownership, not a lack of ambition.
Key Terms
Functional failure
Inability to perform a required function at its specified standard. Defined before failure data is collected or compared.
Simpler comparator
The existing method a model would actually improve: a condition limit, engineering rule, specialist review or better-maintained instrument.
Lead time
How far ahead of the event a warning arrives. Useful only against the time the response actually needs.
Alert playbook
Agreed initial checks, escalation authority, review deadline, permitted interim action and recording requirements when an alert arrives.
Self-Check — Before You Submit
1. A supplier shows a fictional onshore operator a pump-failure model with high overall accuracy, tested on records that are almost entirely normal operation. What should the asset manager ask for?
- A. A higher accuracy figure on the same records
- B. Detected events, missed events, false alerts and lead time, on definitions the team can reproduce
- C. A larger dataset from the same period
- D. Approval to deploy, because the accuracy is high
B. Detected events, missed events, false alerts and lead time, on definitions the team can reproduce — Where normal operation dominates, a system predicting no failures can look accurate while giving little useful warning. A larger dataset is tempting, but more normal records do not reveal missed events, false alerts or lead time.
2. In the book's fictional validation exercise, what fraction of flagged windows contained the target event?
- A. 75%
- B. 82.5%
- C. 16.7%
- D. 33.3%
D. 33.3% — Nine of the 27 flagged windows (9 events plus 18 normal) contained the event: 33.3%. The trap is 75%, which answers a different question: how many of the 12 event windows were detected.
3. A validated model can warn of a developing fault two days ahead. The required replacement has a long manufacturing and mobilisation cycle, and no interim action is defined. What is the right executive conclusion?
- A. The constraint sits in execution: define the response path first
- B. Expand the model to more machines
- C. Ask the supplier for more lead time
- D. Remove the protective trip, since the model warns first
A. The constraint sits in execution: define the response path first — Lead time is useful only in relation to the response; more detection can leave the organisation no better able to act. Expanding the model is the tempting error, and protective systems always stay independent of an advisory.
Today’s Assignment
Write Section 6 of your plan: the reliability case for your Session 1 decision.
- Name the target failure mode in your company's decision, and define the event, the alert and a successful intervention so your team could reproduce them.
- Name the credible simpler comparator your company relies on today, and what false alerts would cost in specialist time.
- State the measures that replace a single accuracy figure (detected events, missed events, false alerts, lead time) and how validation would respect time and operating regimes.
- Write the alert playbook and the response window, and confirm that protective systems and emergency procedures stay independent of any advisory.
- The honest part: where your records say repaired but not what was found, or where your team could not have acted on a warning in time. If the failure mode gives no useful warning, say so plainly: a prediction programme is then the wrong investment. If no AI is in use yet, that is normal. Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 6 — The target failure mode, the simpler comparator and the response window.
How this is marked — the rubric
| Defines the failure mode, event, alert and successful intervention reproducibly | 25 |
| Names the simpler method actually in use as the comparator, with the cost of false alerts | 20 |
| Uses the four measures and validates across time and operating regimes | 25 |
| Writes the playbook and response window, keeps protection independent and states honestly whether prediction is worth funding | 30 |
| Total · pass mark 70 | 100 |
Submitting session 6
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 6, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.Capabilities Covered
Failure-mode definitionComparator disciplineReading validation resultsTime-aware validationAlert playbooks
Chapter 7 · Intelligence in the FieldSESSION 7
Pipelines and the Responsibility to Contain
Why This Matters
Session 6 asked whether a warning can become an action. Pipelines test that question across distance. A line can look simple in the asset register: a route, a diameter, a service and a capacity. The ownership problem is distributed. Sections differ in condition, access, protection and exposure to interference, and a whole-line average can hide the location that most needs attention. This is also sensitive ground. An ambiguous signal is not proof of theft, leakage or equipment failure, and a flag never establishes who is responsible. If your business has no pipelines, apply today to the containment boundary nearest your decision: flowlines, transfer lines, process piping, or the pipeline your business depends on.
25-Minute Module
0:00 – 5:00
Segments and locations, not averages
- Begin with a segment and system view that competent pipeline specialists accept. Connect inspection, operating conditions, repairs, route information and incidents to the appropriate locations.
- An average condition for a whole line is like an average across a facility: it can be comfortable while one location is not. Keep the relationship between the line and its connected facilities visible, because a local condition can have consequences well beyond its physical length.
- Location quality matters. If an inspection finding cannot be reliably linked to the route, later work may examine the wrong place or fail to demonstrate that a previous anomaly was addressed. This is Session 2's identity problem stretched along a route.
- THE DOCTRINE
A precise coordinate on a screen does not mean the underlying observation was equally precise. Preserve the location basis and its uncertainty.
Ask where, exactly, the finding is, and how anyone knows.
5:00 – 10:00
Separate the mechanisms
- Internal deterioration, external corrosion, coating damage, mechanical damage, ground movement and third-party interference are not interchangeable explanations. The evidence needed to investigate each differs, and a model trained to find an operating anomaly does not automatically identify its physical cause.
- Ask the technical team to state the mechanism under consideration and the evidence supporting it. Operational changes, measurement errors and transient conditions can also affect a signal.
- THE DOCTRINE
Do not let an ambiguous pressure change become a confident claim of theft, leakage or equipment failure. The response should be proportionate and governed by the relevant procedures.
- Hold the same line about people: a signal pointing towards a community, contractor or employee is not a finding about them. Keep security-sensitive detail within controlled channels, and obtain appropriate review before combining surveillance information with personal or community-related data.
When someone names a cause, ask which mechanism, and what evidence separates it from the others.
10:00 – 15:00
Surveillance supports a response arrangement
- AI can help compare observations from different sources and direct attention. It can also create false confidence when several sources repeat the same initial report. Two messages forwarding one photograph are not independent corroboration.
- Build source relationships into the evidence trail: a direct observation, an inference, a forwarded report and an unverified statement are different things.
- Detection should connect to an authorised response: operational verification, field inspection, isolation assessment, security escalation or another specified action. Do not create an informal command path in which a software alert instructs field personnel to take hazardous action without the required assessment and authority.
- AT THE ASSET IN THE FIELD · THE WORKED CASE
- In the field. A fictional onshore operator's monitoring flags a possible event on a remote segment, and two messages arrive carrying the same photograph. The control room logs one unverified report, not two confirmations, and starts the authorised response its established procedure specifies, such as operational verification or field inspection. Nobody is sent to act on the alert alone.
IN THE BOARDROOM · THE PARALLEL- In the boardroom. The same operator's board is shown a quarter with no detected events and invited to credit the surveillance platform. The correct reply is that an absence of detected events may reflect many things, including the absence of a threat. Report response time, verified coverage and closure of identified vulnerabilities separately from estimated avoided consequences.
Count sources, not messages.
15:00 – 20:00
Repairs carry history; exposure is counted once
- When a repair is completed, the line does not return to an imaginary state with no history. Record the original condition, assessment, repair design, materials, installation evidence, examination, acceptance and future monitoring requirements, tied to the location and any operating restrictions. A visually neat surface does not explain the repair beneath it.
- THE DOCTRINE
AI can retrieve and cross-check the repair dossier. It should not reconstruct missing acceptance as if it existed.
- Review groups of repairs for common lessons. Repeated interventions in one segment may mean local repairs are not addressing a wider deterioration process or service issue. The portfolio decision might then concern replacement or changed duty; it cannot be settled by ranking individual work orders.
Do not confuse With Why it matters Production deferred Sales permanently lost Different consequences; adding both overlaps Gross throughput The owner's net entitlement Only the owner's share is the owner's cash A transport tariff The commodity value flowing through the line Different economic interests - Use the relevant commercial boundary. For a transport owner, the benefit may relate to service availability and applicable charges; for a producer, to saleable output, timing and contractual arrangements. Finance should identify the cash flow the intervention can actually change.
If the same loss appears in two rows of a business case, it has been counted twice.
20:00 – 25:00
Exercise the response, and your wrap
- Ask for a decision map of the most material segments: condition evidence, important dependencies, accepted restrictions, unresolved anomalies, access readiness and the next decision. It should lead to a working record with sources, not substitute for it.
- Test the response arrangement through an exercise that includes a plausible false alert, a failed communication link and an event whose location is uncertain. Observe whether the team follows established authority or improvises around the software. Correct the workflow before expanding the technology.
- THE DOCTRINE
THE LEADERSHIP CHARGE. Make every material containment concern traceable to a location, a competent assessment and a current disposition. Require a response system behind the surveillance system.
- Before you close: name the containment concern in or nearest to your Session 1 decision, its location and its current disposition. That is Section 7 of your 90-Day Intelligent Asset Plan. Write your assignment now.
An exercise that surprises nobody has not tested the response.
Key Terms
Segment and system view
Condition, repairs, route information and incidents tied to specific pipeline locations and connected facilities, rather than a whole-line average.
Independent corroboration
Confirmation from a separate source. A forwarded copy of the same report is not independent.
Disposition
The authorised treatment of a condition, including any limits, work, monitoring or reassessment requirements.
Commercial boundary
The economic interest an intervention can actually change: for example tariff and availability for a transporter, saleable output and timing for a producer.
Self-Check — Before You Submit
1. A fictional pipeline owner's analytics flag a pressure change on a segment near a community. What should the operations manager do first?
- A. Log suspected theft and name the community in the incident report
- B. Treat the signal as unverified and start the authorised response the relevant procedures specify, such as operational verification
- C. Instruct the nearest field crew to isolate the segment on the strength of the alert
- D. Ignore it until a second alert arrives
B. Treat the signal as unverified and start the authorised response the relevant procedures specify, such as operational verification — An ambiguous pressure change is not proof of theft, leakage or failure, and a flag never establishes who is responsible. Sending a crew to isolate on the alert alone is the tempting error: it creates an informal command path that bypasses the required assessment and authority.
2. Two field messages arrive, each forwarding the same photograph of a suspected leak. How should the evidence trail record them?
- A. As two independent confirmations
- B. As a confirmed leak, because the photograph is clear
- C. As one forwarded report, kept unverified until a direct observation
- D. As a security matter outside the evidence trail
C. As one forwarded report, kept unverified until a direct observation — Two messages forwarding one photograph are not independent corroboration. Counting them twice is the trap; a clear image still documents appearance, not a verified cause.
3. A producer's business case for a line repair adds production deferred, sales permanently lost and the full commodity value flowing through the line. What is wrong?
- A. It adds overlapping quantities outside the producer's commercial boundary; finance should identify the cash flow the repair can actually change
- B. Nothing, provided the figures come from the finance team
- C. It should also add the transport tariff
- D. It should use gross throughput instead
A. It adds overlapping quantities outside the producer's commercial boundary; finance should identify the cash flow the repair can actually change — Deferred production and permanent loss are different and must not be added together, and the full value flowing through the line is gross throughput, not the producer's net entitlement. Adding the tariff or using gross throughput repeats the same error of counting interests the producer does not hold.
Today’s Assignment
Write Section 7 of your plan: the containment concern in or nearest to your Session 1 decision, traced to location, assessment and disposition, with the response behind any surveillance.
- Trace the containment concern in or nearest to your company's decision to a location (with its basis and uncertainty), the mechanism under consideration and its evidence, the competent assessment and its current disposition.
- Record the repair history it carries: original condition, assessment, repair design, materials, installation evidence, examination, acceptance and monitoring. Name what is missing.
- Describe the authorised response arrangement behind any surveillance: who verifies, who decides, which actions are permitted, how sources are recorded, and how security-sensitive and personal or community-related information is handled.
- State the commercial boundary and the cash flow the intervention can actually change, counted once.
- Design the exercise: a plausible false alert, a failed communication link and an uncertain location. Say what you would observe.
- The honest part: where your business currently treats a signal as proof, or a quiet period as evidence of success. If your business has no pipelines, use the nearest containment boundary or the line you depend on. If no AI is in use yet, that is normal. Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 7 — Containment traced to location, assessment and disposition, with a response behind the surveillance.
How this is marked — the rubric
| Traces the concern to a location with its basis and uncertainty, a stated mechanism, a competent assessment and a current disposition | 30 |
| Records the repair history and names the missing elements without reconstructing them | 20 |
| Describes an authorised response with source discipline and review of sensitive data, attributing no blame from a signal | 20 |
| States the commercial boundary and changeable cash flow, counted once | 15 |
| Designs a response exercise with all three test conditions, and gives an honest example | 15 |
| Total · pass mark 70 | 100 |
Submitting session 7
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 7, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.Capabilities Covered
Segment-level ownershipMechanism disciplineResponse arrangementsRepair lifecycle recordsCommercial boundaries
Chapters 8 and 9 · Intelligence in the FieldSESSION 8
Offshore Integrity and the Cost of Access
Why This Matters
Today you follow the work to places where access shapes everything, and through the systems between the large machines. An offshore inspection or repair is not just an engineering scope transported to sea, and a facility's usable capacity can be limited by a utility, a protective function, a transfer line or a tank rather than its most expensive unit. The least expensive mobilisation is not necessarily the least expensive accepted job. Nor is a lower trip count automatically better performance. If your assets are onshore, apply the offshore half to the most access-constrained work in your decision: a swamp location, a remote site or work inside an operating facility.
25-Minute Module
0:00 – 5:00
The system behind the structure, and access as an input
- Access, movement, lifting, weather, accommodation, simultaneous operations and emergency arrangements shape what is feasible; represent them early enough to influence the decision. Pressure to use an available window is precisely when unclear scope and weak acceptance criteria are expensive.
- The visible steelwork is only part of the question: connections, moorings, marine systems, pressure equipment, electrical and control equipment, utilities and protective arrangements may matter, as the competent disciplines decide. A surface image documents appearance; it cannot on its own establish fatigue condition, hidden deterioration or mooring adequacy.
- Compare access options (rope access, scaffolding, robotic systems, remotely operated underwater equipment, vessel-supported methods) against the actual task. Count preparation, permits, competence, reach, recovery and unusable evidence.
- On an operating floating asset, verify the motion and interface assumptions: a demonstration on a stationary surface does not establish performance on a moving hull. Require evidence for the intended conditions and a defined fallback.
Ask which parts of the technical conclusion rest on measurement, analysis, assumption or areas nobody reached.
5:00 – 10:00
Coating purpose, campaign coupling and the field voice
- A repainting programme should not conceal uncertainty about the substrate. Define inspection before and during preparation, escalation of unexpected conditions and acceptance evidence before release. An assumed daily rate is not a fact about an installation, and the handover record must distinguish treated, accepted and inaccessible areas.
- Campaign coupling means individually reasonable jobs competing for the same access, isolation, lifting support or specialist. AI can analyse those shared constraints; the responsible teams review the suggested sequence. Where a preliminary inspection could reveal substantial extra work, reserve a decision point and contingency, and know what finding changes the scope and who authorises the response.
- A nominally mobilised team may still lack a competence, certificate, tool or approved procedure; check readiness against the task.
- Remote leadership sees the summary; the field team sees the surface, the movement and the limits. Give the field team a route to report that an approved method does not fit without being treated as an obstacle, and route that report to a decision, not a reassuring status update.
- THE DOCTRINE
THE LEADERSHIP CHARGE (Chapter 8). Approve the evidence, access and contingency plan together. Offshore productivity has little meaning until the work can be performed, accepted and handed back under the actual conditions.
Ask supervisors what the remote team routinely misunderstands.
10:00 – 15:00
Utilities and protective functions
- Follow the required service: for a refinery, a train and its supporting services; for a gas facility, compression, treatment and delivery conditions; for a terminal, receiving, storing and dispatching product within the accepted configuration.
- A utility can support several units, so its failure creates consequences no single equipment business case shows. Establish the functions served, accepted redundancy and restoration arrangements. AI can suggest dependencies from drawings, but text proximity is not proof of a physical connection; keep the accepted drawing revision behind the map.
- Protective functions are not ordinary production assets. IEC 61511-1's published scope covers specification, design, installation, operation and maintenance of safety instrumented systems for the process sector. An advisory analytics application does not inherit that qualification because it uses the same process data.
- THE DOCTRINE
An AI recommendation to reduce nuisance trips is not authority to change a setpoint, bypass an interlock or extend a test interval. Such changes require the applicable engineering, operational and regulatory processes.
- Ask whether a proposed availability improvement removes a genuine defect or weakens a protection that is doing its job. Record causes and dispositions so a favourable trend does not conceal a degraded safeguard.
A digital twin becomes a liability when nobody knows which configuration it represents.
15:00 – 20:00
Storage, process change and one facility review
- A tank's condition and supporting arrangements determine whether its capacity is usable; the specialist sets the examination and acceptance basis. An acceptable sample is not evidence about every inaccessible location, and a change of stored product can affect technical suitability, not just commercial allocation.
- A change of feed, throughput or regime can make a model trained on earlier conditions less relevant. Ask what change requires reassessment, and record interventions and operating changes: an improvement after maintenance may coincide with a lower load.
- AT THE ASSET IN THE FIELD · THE WORKED CASE
- In the field. A fictional product-terminal owner plans to take one tank out of service for examination. The maintenance planner sees a contained job. Operations sees that the remaining tanks and transfer path may now carry more of the product movement plan, which can concentrate work on them.
IN THE BOARDROOM · THE PARALLEL- In the boardroom. The same owner's finance team is shown the tank repair as added storage revenue. The benefit depends on demand, the availability of connecting transfer equipment and the commercial ability to use the restored capacity. A theoretical increase in storage availability is not automatically additional revenue. One tank, read from two ends of the business.
- Bring the next production, maintenance and inventory plans into one discussion. Ask which availability assumptions each depends on, and resolve contradictions while there is time to change the sequence or fund a missing prerequisite. A dependable conflict check can be more useful than a sophisticated failure model that leaves the plans disconnected.
Look for the tank, line or utility that two plans both assume is free.
20:00 – 25:00
The charge, and your wrap
- THE DOCTRINE
THE LEADERSHIP CHARGE (Chapter 9). Follow the required service through its supporting systems. Protect the distinction between production optimisation and the independent arrangements that make operation acceptable.
- Before you close: name the most access-constrained work in your Session 1 decision and the supporting system your required service depends on. That is Section 8 of your 90-Day Intelligent Asset Plan.
- Pop Quiz 2 opens now: five questions, closed book, on Sessions 6 to 8. Sit it before Session 9, then write your assignment.
The cheapest quotation can be the one that has not yet met the conditions.
Key Terms
Accepted job
Work performed, examined, accepted and handed back under actual conditions. The basis for comparing access options, rather than mobilisation cost.
Campaign coupling
Separate jobs in one campaign competing for the same access, isolation, lifting support or specialist.
Safety instrumented system
A protective function with its own lifecycle discipline; IEC 61511-1's published scope covers specification, design, installation, operation and maintenance in the process sector.
Required service
The duty the facility must deliver, followed through the utilities, protection, transfer and storage it depends on.
Self-Check — Before You Submit
1. A contractor offers the lowest mobilisation cost for an inspection on an operating floating asset, using a robotic system demonstrated on a stationary surface. What should the owner require?
- A. Evidence relevant to the motion and interface conditions, and a defined fallback
- B. Nothing further, because the price is lowest
- C. A larger daily coverage rate in the quotation
- D. A demonstration on another stationary surface
A. Evidence relevant to the motion and interface conditions, and a defined fallback — A stationary demonstration does not establish performance on a moving hull, and the cheapest mobilisation is not necessarily the cheapest accepted job. A higher coverage rate is tempting, but an assumed rate is not a fact about the installation.
2. Which statement about the benefit of a tank repair at a terminal is correct?
- A. Restored capacity always becomes revenue
- B. The benefit equals the tank's capacity multiplied by the storage charge
- C. The benefit is an engineering question only
- D. The benefit depends on demand, connecting transfer equipment and the commercial ability to use the capacity
D. The benefit depends on demand, connecting transfer equipment and the commercial ability to use the capacity — A theoretical increase in storage availability is not automatically additional revenue. The multiplication answer is tempting because it looks precise, but it ignores whether the transfer path is available and whether the capacity can be sold.
3. An analytics tool recommends raising a trip setpoint on a compressor to reduce nuisance trips. What should the operations director do?
- A. Approve it, because availability will improve
- B. Let the tool adjust the setpoint within a small band
- C. Route it through the applicable engineering, operational and regulatory processes
- D. Bypass the interlock until the cause is known
C. Route it through the applicable engineering, operational and regulatory processes — An AI recommendation is not authority to change a setpoint, bypass an interlock or extend a test interval. A small automatic band is the tempting compromise, but it still lets an advisory tool alter a protective function.
Today’s Assignment
Write Section 8 of your plan: access, evidence and contingency planned together for your Session 1 decision, and the required service followed through its supporting systems.
- Name the most access-constrained work in your company's decision and compare realistic access options on the accepted job, not the mobilisation.
- Identify the campaign coupling, the decision point and contingency for discoveries, and who can authorise a change of scope.
- Describe how your field team can report that an approved method does not fit the conditions, and how that report reaches a decision.
- Follow the required service through its utilities, protective functions (and who controls their basis and testing), storage and the process changes that would require reassessment.
- State the contradiction between your production, maintenance and inventory plans that must be resolved first, and who resolves it.
- The honest part: where your business has judged access on mobilisation price, or read a falling trip count as good news without asking why. If no AI is in use yet, that is normal. Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 8 — Access, evidence and contingency planned together, and the required service followed through its supporting systems.
How this is marked — the rubric
| Compares access options on the accepted job, with coupling, a decision point and contingency authority named | 30 |
| Gives the field team a defined route to report a method that does not fit, ending in a decision | 15 |
| Follows the required service through utilities, protection, storage and process change, keeping protective functions outside advisory authority | 30 |
| Names the contradiction between the three plans with an owner by role, and gives an honest example | 25 |
| Total · pass mark 70 | 100 |
Submitting session 8
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 8, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.ASSESSMENT NOW OPEN
Pop Quiz 2 — Failure Modes, Containment and Access opens now. Sit it before you start the next session — it is how you find out whether the doctrine landed, and the practice papers may be resat as often as you like.
Sit Pop Quiz 2 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Access planningCampaign couplingProtective-function governanceWhole-function storage viewCross-plan conflict checks
Chapter 10 · Lifecycle ChoicesSESSION 9
Repair, Replace or Operate with Restrictions
Why This Matters
Your decision may now be close to an intervention. Today is about choosing it properly. I would ask you to slow down here. A repair proposal can become persuasive before the engineering question is settled: it offers a price, a duration and a return to normality. A replacement can be equally attractive because it looks definitive. Leadership needs to establish what each option actually resolves and what it leaves behind. We will put eligibility before price, keep approvals separate, date every temporary repair, treat restriction as an active arrangement and see a replacement's interfaces. None of this is an engineering assessment.
25-Minute Module
0:00 – 6:00
Feasibility comes before ranking
- Begin with the condition and the required future duty: the mechanism, extent, relevant loads, proposed operating period and uncertainty. The technical authority determines which options are eligible for comparison. Only then do procurement and finance compare technically feasible choices on a consistent basis. This is the Value-versus-Risk Map: engineering acceptance is a gate, not a weight.
Approval What it establishes Engineering assessment That a repair concept is suitable within defined conditions Commercial authority That the cost is approved; it does not certify fitness for service Operations That the work may proceed within an accepted isolation and access arrangement Examination and acceptance That the completed work may return to the intended service A model's recommendation None of the above; it cannot supply any missing authority - THE DOCTRINE
These decisions should connect without collapsing into one signature. A purchase approval does not certify fitness for service, and a contractor's availability does not establish that the repair addresses the damage.
- Assessment and repair are different disciplines. Fitness for service is an engineering assessment of suitability for specified service. API's catalogue describes API 579-1/ASME FFS-1 for assessment, while ASME PCC-2 addresses repair of in-service pressure equipment and piping. Require competent selection of the applicable method and edition; mentioning a standard does not establish compliance.
Ask who decided this option was eligible, before anyone tells you what it costs.
6:00 – 11:00
Temporary needs a date and conditions
- A repair called temporary can stay in place because its replacement keeps moving to the next budget. Record the accepted service, monitoring, inspection, expiry or reassessment conditions and the planned permanent disposition where one is required. The exact requirements come from the engineering and regulatory basis for the job.
- THE DOCTRINE
Treat expiry as a decision trigger, not an administrative inconvenience. Changing the due date in a maintenance system does not renew the technical assessment.
- AI can identify approaching deadlines, locate the acceptance record and flag inconsistency between a repair record and the current operating service. It should not extend the acceptance period automatically or assume a previous repair remains suitable after the duty changes.
- Specialist repairs, composite repairs among them, can suit defined applications. Suitability depends on the actual damage, geometry, material, service, loading, installation conditions and qualification; a product's general capability is not acceptance for a particular component. Ask for the design basis, installation procedure, competence requirements, quality evidence and future inspection arrangement. Welded and mechanical repairs and replacement spools need the same controlled basis.
- Compare complete accepted scopes, including preparation, access, testing, monitoring and future obligations. A low initial quotation can omit work you will later have to fund.
Ask when each temporary repair expires, and what decision is booked for that date.
11:00 – 16:00
Restricted operation is an active arrangement
- In plain language, the operating envelope is the defined set of conditions within which operation is authorised. A restriction narrows it, and someone has to keep it narrowed. Define the permitted envelope, the means of verifying it, the person monitoring it and the response if conditions move outside the accepted basis.
- A restriction that depends on a measurement needs that measurement available and reliable; one that depends on a manual task needs staffing, workload and handover addressed. A restriction written in a report but absent from operating instructions is not an effective operating arrangement.
- AT THE ASSET IN THE FIELD · THE WORKED CASE
- In the field. A fictional swamp-location operator's technical authority has accepted continued service of a flowline under a restriction while a permanent repair is designed. The limit sits in the operating instructions, the shift can read the governing measurement, a named person monitors it and the response to an excursion is written down. An advisory model may help detect a possible departure.
IN THE BOARDROOM · THE PARALLEL- In the boardroom. The same operator's executive committee is asked to fund the permanent repair. The paper must say what the limit is, who verifies it, when it is reassessed and what happens if the repair slips. The committee holds spending authority; it neither sets nor relaxes the restriction. Same decision, seen from the shift and from the funding table.
- THE DOCTRINE
Do not let AI become the only custodian of the limit. The approved operating controls should remain usable when the advisory application is unavailable.
Ask the night shift how they would know the restriction had been breached.
16:00 – 20:00
Replacement has interfaces too
- New equipment can introduce changed loads, control behaviour, power demand, maintenance needs or spare requirements. It should not be treated as identical merely because its purchase description looks similar. Apply the relevant management-of-change and commissioning processes.
- Include removal, installation, integration, training, initial spares and acceptance in cost and schedule. A delivered item is not an operating asset. See the time between payment, delivery, installation and demonstrated service, especially when the financial case assumes benefits immediately after purchase.
- Retain the replaced equipment's history and its relationship to the new configuration, so later analysis can tell a change in condition from a change in identity (Session 2) and model comparisons stay honest.
Ask on what date the replacement is expected to deliver demonstrated service, not on what date it arrives.
20:00 – 25:00
The decision paper, and your wrap
- The recommendation should identify the eligible options, the engineering basis for each, lifecycle costs, execution constraints, residual obligations and the limits of the requested approval. If further examination is needed before a repair can be designed, approve that investigation explicitly rather than implying the full intervention has been sanctioned.
- Preserve the rejected options and reasons. That helps when circumstances change and stops a later team reopening a resolved argument without its evidence.
- THE LEADERSHIP CHARGE. Require an accepted technical scope before comparing prices. Make restrictions, reassessment conditions and return-to-service evidence part of the decision from the beginning.
- Before you close: list the eligible options for your Session 1 decision and the approval each one still needs. That is Section 9 of your 90-Day Intelligent Asset Plan. Write your assignment now.
An approval that does not state its own limits will be read as approval of everything.
Key Terms
Fitness for service
An engineering assessment of suitability for specified service. A model prediction or management score is not such an assessment.
Technical authority
A person or function authorised, within defined competence and delegation, to make specified engineering decisions, including which options are eligible.
Operating envelope
The defined conditions within which operation is authorised, governed through the applicable engineering and operating arrangements.
Temporary repair
A repair accepted for a stated service with monitoring, expiry or reassessment conditions and, where required, a planned permanent disposition.
Self-Check — Before You Submit
1. A temporary repair is approaching its reassessment date. The planner proposes moving the due date in the maintenance system because nothing appears to have changed. What should the asset manager do?
- A. Approve the new date, since the condition looks stable
- B. Let the AI scheduling tool extend it automatically
- C. Ask the contractor to confirm it is still fine
- D. Treat the date as a decision trigger and obtain a new authorised disposition before continuing on that basis
D. Treat the date as a decision trigger and obtain a new authorised disposition before continuing on that basis — Changing the due date in a maintenance system does not renew the technical assessment, and AI should not extend acceptance automatically. A stable appearance is the tempting reason, but it is not an assessment.
2. A repair proposal states that the work will follow ASME PCC-2. What does that statement establish on its own?
- A. That the equipment is fit for service
- B. Not compliance: the owner still requires competent selection of the applicable method and edition
- C. That API 579-1/ASME FFS-1 is no longer needed
- D. That the commercial approval can proceed without engineering review
B. Not compliance: the owner still requires competent selection of the applicable method and edition — ASME PCC-2 addresses repair; API 579-1/ASME FFS-1 is described for assessment, and they are different disciplines. The trap is reading a standard's name as proof: mentioning a standard does not establish compliance or fitness for service.
3. Which item belongs in the executive decision paper for a repair, replace or restrict choice?
- A. Only the preferred option
- B. The contractor's capability statement instead of the engineering basis
- C. The limits of the approval requested and the rejected options with reasons
- D. A single weighted score settling safety and cost together
C. The limits of the approval requested and the rejected options with reasons — The paper states the limits of the requested approval and preserves rejected options with reasons. A single weighted score is tempting, but engineering acceptance is a gate that cost cannot outweigh.
Today’s Assignment
Write Section 9 of your plan: the repair, replace or restrict decision for the asset in your Session 1 decision.
- Set out the eligible options for your company's asset and who determines eligibility, by role. State the accepted technical scope you would require before any price is compared.
- For each option, list the separate approvals it needs: engineering, commercial, operations, and examination and acceptance.
- For any temporary repair, state the date and conditions attached and the decision booked at expiry.
- For any restriction, state the envelope, how it is verified and by whom, how it is enforced, and how it survives without any advisory tool.
- For any replacement, name the interfaces it introduces and the time to demonstrated service. Then write the contents of the executive decision paper, including the limits of the approval requested and the rejected options with reasons.
- The honest part: where your company has a temporary repair or restriction that nobody now owns. If no AI is in use yet, that is normal. Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 9 — Eligible options, the accepted technical scope before price, and restrictions with reassessment conditions.
How this is marked — the rubric
| Names eligible options and the role that determines eligibility, with an accepted technical scope required before price | 25 |
| Keeps engineering, commercial, operations and acceptance approvals distinct for each option | 25 |
| Attaches a date and conditions to any temporary repair, and makes any restriction verifiable and enforceable without an advisory tool | 25 |
| States replacement interfaces and a decision paper with approval limits and rejected options, plus an honest example | 25 |
| Total · pass mark 70 | 100 |
Submitting session 9
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 9, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.Capabilities Covered
Eligibility before priceSeparated approvalsTemporary-repair controlRestriction ownershipReplacement interfaces
Chapter 11 · Lifecycle ChoicesSESSION 10
Shutdown Readiness Is More Than a Date
Why This Matters
Many asset decisions end up waiting for a window: a shutdown, a turnaround, a campaign. A shutdown date can be fixed long before the work is ready, and that creates a familiar temptation to report progress against the date while assuming the missing scope, material or decision will arrive in time. The owner should govern readiness through evidence, not through the colour expected at a meeting. Today you will sort the work that belongs in the window, separate the readiness clocks, find what would really move the return-to-service date, treat restart as its own decision and close Week 2.
25-Minute Module
0:00 – 5:00
The work that belongs in the window
- Bring the equipment condition and the intervention decisions from Sessions 3 to 9 into scope selection. Identify four categories: mandatory work, work justified by condition, opportunity work and discretionary improvement. Keep them visible so that commercial pressure does not quietly remove work necessary for the operating basis.
- Each work package needs enough definition to plan access, isolation, materials, competence, examination and handback. A short work order title is not a prepared package. Link each package to the accepted scope and the evidence that will demonstrate completion.
- Scope freeze is the point after which additions and deletions pass through a controlled route. Discoveries can justify change; the freeze is not there to deny reality but to expose a change's effect on resources, sequence, duration and risk. Whoever approves added work should see what other work or contingency it consumes.
Ask which mandatory item was removed last time, and who noticed.
5:00 – 11:00
Readiness has several clocks
- In plain language, readiness is evidence that the prerequisites for a specified task have been met. It is distinct from the task's priority: Session 4 already kept exposure and readiness apart. And readiness runs on several clocks at once:
This clock can show ready While this one is not Engineering Procurement is late Materials delivered Certificates remain unaccepted Contractor available Site access or task-specific competence is unresolved Isolation plan prepared A connected operation has not agreed the interface - THE DOCTRINE
A combined green status can hide a single failed prerequisite that prevents the work starting. A forecast date should come from the dependencies, not from the colour expected at a meeting.
- Show the conditions separately. Use necessary gates for critical prerequisites, with a clear escalation route when a gate is not met.
- AI can cross-check work packages against material status, required documents and contractor records, and find evidence missing from an apparently complete package. Human planners and technical reviewers establish whether each prerequisite is applicable and adequate.
When every line on the readiness report is green, ask which single item would stop the work tomorrow.
11:00 – 16:00
The critical path is a work relationship
- The critical path is the chain of linked activities that decides when the work can finish. A list of activities sorted by date does not establish it; the schedule needs logical relationships, resource constraints and durations appropriate to the scope.
- Ask the planning team which unfinished activity would move the return-to-service date, and what assumption supports that conclusion.
- Use scenario analysis for the important uncertainties: a delayed specialist, a failed acceptance test, a discovered condition. A scenario should help you decide whether to add resources, change scope, move the window or prepare a contingency, not produce a wider range around the same optimistic date.
- Preserve the difference between a model forecast and an approved plan. If AI predicts a delay, route it to the planner and the decision owner. Do not rewrite the baseline invisibly; the organisation must know whether it is seeing a warning, an updated forecast or a sanctioned change.
A forecast that silently becomes the plan has skipped the decision.
16:00 – 21:00
Restart, and a fictional readiness review
- THE DOCTRINE
Mechanical completion, testing, reinstatement and operational readiness are different states. Know what evidence must be accepted before a system returns to service, and who has the authority to release it.
- Unfinished work needs a disposition: some items may be accepted for later completion under defined conditions; others prevent startup. Specialists set the categories, and an AI summary must not describe all outstanding work as minor punch items. Changed training, drawings and maintenance tasks belong in the completion account.
- AT THE ASSET IN THE FIELD · THE WORKED CASE
- In the field. The book's fictional readiness review: a campaign is approaching and its main replacement item has arrived. The team reports the package ready. A document check shows the material traceability record is incomplete and the receiving examination has not been accepted. The schedule assumes immediate installation.
IN THE BOARDROOM · THE PARALLEL- In the boardroom. The owner has a decision, not a reporting problem. Can the evidence be obtained and accepted in time? Is a technically approved alternative available? Does the window need to change? Reclassifying the package as ready resolves nothing. AI can connect the missing acceptance to the scheduled installation and alert the people who can act; it must not invent the certificate, waive the requirement or bury the issue beneath a revised forecast.
- The right answer depends on the equipment and the evidence, so it belongs with the people who hold that authority.
When a package is reported ready, ask to see the acceptance, not the delivery note.
21:00 – 25:00
Learn from the campaign, and close Week 2
- Record what drove the final duration and cost: scope discovery, waiting, access, rework, testing or another cause. Distinguish planned contingency from avoidable delay, and compare the forecast with actual work without changing the original definitions afterwards.
- Return the findings to the equipment history, planning assumptions and supplier assessments. The Operating-Intelligence Loop closes through this learning, not through a celebratory completion slide.
- THE DOCTRINE
THE LEADERSHIP CHARGE. Ask which unaccepted prerequisite can still prevent the critical work or restart. Resolve it through evidence and authority before the window opens.
- Before you close, I would like you to name the window your Session 1 decision depends on and its weakest readiness clock. That is Section 10 of your 90-Day Intelligent Asset Plan, and it completes Week 2.
- Practice Exam 2 opens now: fifteen questions, closed book, on Sessions 1 to 10, closing Week 2. Sit it before Session 11, then write your assignment.
If the only lesson recorded is that the shutdown finished late, the next one has nothing to learn from.
Key Terms
Scope freeze
The point after which additions and deletions to the window's work pass through a controlled route that shows their effect on resources, sequence, duration and risk.
Readiness
Evidence that the prerequisites for a specified task or decision have been met. Distinct from the task's priority.
Critical path
The linked activities that set the return-to-service date, established by logical relationships, resource constraints and realistic durations, not by sorting dates.
Release for return to service
The authorised decision, on accepted evidence, that a system may restart. Distinct from mechanical completion, testing and reinstatement.
Self-Check — Before You Submit
1. At a fictional readiness review, the main replacement item has arrived but its material traceability record is incomplete and the receiving examination has not been accepted. The schedule assumes immediate installation. What should the owner do?
- A. Report the package as ready, since the item is on site
- B. Treat it as an owner decision: can the evidence be accepted in time, is an approved alternative available, or must the window move
- C. Ask the AI tool to generate the missing certificate from supplier data
- D. Install now and complete the paperwork after restart
B. Treat it as an owner decision: can the evidence be accepted in time, is an approved alternative available, or must the window move — An incomplete traceability record and an unaccepted examination are an owner decision, not a reporting problem. Reclassifying the package as ready is the tempting shortcut; AI must not invent the certificate, waive the requirement or bury the issue.
2. A shutdown dashboard shows one combined green status. Engineering and the contractor are ready, but certificates for delivered materials remain unaccepted. What should the asset director require?
- A. Show each readiness clock separately, with a necessary gate for the critical prerequisite and an escalation route
- B. Accept the green status, because most clocks are ready
- C. Add a contingency day to the forecast date
- D. Move the certificate check to after restart
A. Show each readiness clock separately, with a necessary gate for the critical prerequisite and an escalation route — A combined green status can hide a single failed prerequisite. Adding a day is the tempting fix, but a forecast date should come from the dependencies, not padding.
3. An AI scheduling tool predicts a delay to the critical work. What should happen next?
- A. The tool updates the baseline so the plan stays current
- B. The prediction is ignored until the delay materialises
- C. Route it to the planner and the decision owner, and keep it distinct from the approved plan
- D. The return-to-service date is moved automatically
C. Route it to the planner and the decision owner, and keep it distinct from the approved plan — A model forecast is not an approved plan. Updating the baseline automatically is the tempting error: nobody then knows whether they see a warning, an updated forecast or a sanctioned change.
Today’s Assignment
Write Section 10 of your plan: readiness by evidence for the work window behind your Session 1 decision.
- Name the window your company's decision depends on and the work that belongs in it, by category: mandatory, justified by condition, opportunity and discretionary. Say how scope changes after freeze are controlled and who sees what they consume.
- Set out the separate readiness clocks for the critical work, and the necessary gates for critical prerequisites, with the escalation route.
- Name the activity and the assumption that would move the return-to-service date, and one scenario you would analyse.
- State the restart evidence, the states it must pass through and who releases the system to service.
- Name the unaccepted prerequisite that could still prevent the work or the restart, and how you will resolve it through evidence and authority before the window opens. Say how the actual campaign's learning returns to the equipment history.
- The honest part: where your company has reported a package ready on a combined status that hid a failed prerequisite. If no AI is in use yet, that is normal. Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 10 — Readiness by evidence: the window's work, its clocks, the critical path and the restart decision.
How this is marked — the rubric
| Sorts the window's work into the four categories with a controlled route for changes after freeze | 20 |
| Shows separate readiness clocks with gates for critical prerequisites and an escalation route | 25 |
| Names the date-moving activity and its assumption, and the restart evidence with the releasing authority by role | 25 |
| Names the unaccepted prerequisite with a plan to resolve it, the learning route to equipment history, and an honest example | 30 |
| Total · pass mark 70 | 100 |
Submitting session 10
Write your answer in Word or Google Docs. The first line must read
Write your answer in Word or Google Docs. The first line must read
Session: 10, or the submission is refused. Upload the Word file (.docx), or switch to the Google Doc tab and paste your document’s share link.Marked against the rubric above and returned as a written letter, saved to your progress page. Do not submit until you have passed the self-check above.ASSESSMENT NOW OPEN
Practice Exam 2 — Field Intelligence, Repair and Shutdown Readiness opens now. Sit it before you start the next session — it is how you find out whether the doctrine landed, and the practice papers may be resat as often as you like.
Sit Practice Exam 2 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Window scope controlReadiness by evidenceCritical-path challengeRestart governanceCampaign learning