30-Session Programme · 25 Minutes Per Day · Free · For leaders in Nigeria’s oil and gas sector
WEEK 4 OF 6 · RELIABILITY, PROJECTS AND DELIVERY · SESSIONS 16–20 · WORK AT YOUR OWN PACE
This week’s assessments: Pop Quiz 4 — Reliability, Integrity and Resilience after Session 18 · Practice Exam 4 — Midpoint after Session 20
By Session 30: you hold a signed 90-Day Intelligent Operating Plan for your own company · Every assignment marked against a published rubric · Pass mark 70
Part Three · Processing, Reliability and Capital DeliverySESSION 16
Reliability, Maintenance and Shutdown Execution
Session 16 of 30 · 25-Minute Module
Why This Matters
It is tempting to start a reliability conversation with the model: how accurate, how early, how many failures it sees. My book starts somewhere less comfortable. A prediction is worth something only if it changes a maintenance decision early enough to prevent an avoidable consequence, and that depends on spares, authorisation, access and the release of equipment as much as on the algorithm. Today you rank equipment by the consequence of losing it, ask for a condition your team can act on, close the chain from warning to confirmed work, and write an investment case that shows its downside. Then the shutdown, where the same discipline meets time pressure.
25-Minute Module
0:00 – 5:00
Start with the failure consequence
Criticality should reflect the consequence of losing the equipment and the actual ability to recover. A machine with a nominal standby may still be critical if the standby is unavailable or unsuitable for the present duty.
THE DOCTRINE
A useful failure prediction must change a maintenance decision early enough to prevent an avoidable consequence. It has little value when the spare is unavailable, the work cannot be authorised or the equipment cannot be released.
Agree the equipment hierarchy and failure definitions before comparing performance. The ISO description of ISO 14224 identifies standardised collection and exchange of reliability and maintenance data for petroleum, petrochemical and natural-gas equipment (ISO, ISO 14224:2016, official standard description). It is a useful reference for consistent records; implementing the detailed standard requires the applicable text and competent review.
Distinguish a defect, a functional failure, a protective trip and a planned intervention. A model trained on inconsistent labels may learn the recording habits of different technicians rather than the failure behaviour of equipment.
Before anyone shows you a failure prediction, ask what the plant calls a failure.
5:00 – 10:00
Predict a condition you can act on
Choose an application where the evidence can support a useful action within the available lead time. Do not begin by promising to predict every failure.
Vibration, temperatures, pressures, electrical signals and maintenance observations may reveal departures from expected behaviour. Their relevance depends on the equipment and failure mechanism, so a competent reliability engineer should define the diagnostic question.
Test false alarms as operational events. An unnecessary intrusive task consumes resources and can itself introduce risk; a missed warning can leave the plant exposed. Evaluation should reflect both consequences.
Every warning should show its basis, time, confidence limits and recommended next step. Where the right response is inspection, say so. A statistical anomaly is not an instruction to dismantle equipment: on the Autonomy Ladder from Session 6, this is a reviewed recommendation.
Accuracy is a question for the model. Lead time and the cost of being wrong are questions for you.
10:00 – 15:00
Close the response chain
Reliability needs a complete chain: detection, diagnosis, planning, materials, access, execution and confirmation. It is the Operating-Intelligence Loop in maintenance terms. A better model improves only the first links.
THE DOCTRINE
Every accepted warning needs an owner and a disposition. Silence should not count as acceptance.
The owner may confirm the concern, collect further evidence, plan work or record a technically justified rejection. Link the warning to the maintenance system but preserve the authorised work process: scope, materials, competence, access and permits. Automatically creating many low-quality tasks can bury critical work in a growing backlog.
Monitor the time between stages. If diagnosis is fast but material approval takes three weeks, the book notes, a more accurate model may leave the principal delay intact. Fund the limiting step: Session 3's binding constraint at the scale of one machine.
AT THE ASSET · THE WORKED CASE
Asset. A fictional onshore operator's monitoring flags a change in a transfer pump. The standby is awaiting repair and cannot carry the duty, so the pump is more critical than its listing suggests. The reliability engineer confirms the concern and recommends inspection.
IN THE OPERATING REVIEW · THE PARALLEL
Review. At the weekly reliability review the same warning is a stage-time problem: diagnosis was quick, the material request is still awaiting approval. The decision is who approves it and whether the standby repair is funded. The same loop, seen from the other end.
Ask for the time between stages, not just the number of alerts. The slowest stage is where the money should go.
15:00 – 22:00
An investment case with a downside, and a shutdown that can be executed
The book's illustrative case, assumed inputs: monitoring at $180,000 up front and $25,000 a year, assumed to prevent four failures a year, each otherwise stopping 600 saleable barrels per day for two days at $25 net contribution per barrel; three years, year-end benefits, 12% discount rate.
Line
Base: four avoided failures
Downside: two avoided failures
Annual gross benefit
4 × 2 × 600 × $25 = $120,000
2 × 2 × 600 × $25 = $60,000
Annual net benefit after $25,000 recurring cost
$95,000
$35,000
Present value (three-year factor approximately 2.401831)
Approximately $228,174
Approximately $84,064
NPV after $180,000 initial cost
Approximately $48,174
Approximately negative $95,936
The recurring cost does not halve with the benefit, so the case turns negative. It also assumes the lost production is not later recovered, the response can prevent the event and no other project claims the benefit. It excludes taxes, financing and terminal value. Those limits stay beside the result.
THE DOCTRINE
A strong predictive score does not establish that four failures will be avoided. The response process and the counterfactual evidence matter equally.
A shutdown tests the same discipline under time pressure. Start from agreed scope and the basis for each task; identify the critical path, long-lead materials, competence, isolations, inspection dependencies and restart conditions. Planners validate any AI-compared schedule against access, simultaneous work and resources.
Give late scope an explicit decision, showing its effect on duration and risk. Assess restart readiness independently against agreed criteria; time pressure does not replace verification. After restart, confirm the intended condition was achieved.
A schedule with impossible crew overlaps is not improved by its attractive graphic.
22:00 – 25:00
Before you close the laptop
Write one line: the equipment decision inside your Session 1 decision whose failure would cost most, and whether its standby is real.
This session produces Section 16 of your 90-Day Intelligent Operating Plan: failure consequence, the actionable condition and the closed response chain.
No process plant? Use the equipment decision you support for an operator, or your own fleet or base equipment. With no model in use, the right answer may be a simpler method, or none.
Leadership earns its return on reliability by ensuring the organisation can act on the evidence.
Key Terms
Criticality
The ranking of equipment by the consequence of losing it and the actual ability to recover, including whether any standby can carry the present duty.
Actionable condition
A predicted condition the evidence supports and the organisation can respond to within the available lead time, with a stated next step.
Response chain
Detection, diagnosis, planning, materials, access, execution and confirmation. A prediction is only as useful as the slowest link.
Warning disposition
The owner's recorded decision on an accepted warning: confirm, collect further evidence, plan work or reject with technical justification. Silence is not a disposition.
Self-Check — Before You Submit
1. A compressor is ranked low criticality because the register shows a standby. What would most change that ranking?
A. Its purchase price is high
B. The standby is unavailable or unsuitable for the present duty
C. A vendor model rates it healthy
D. It has run a year without a trip
B. The standby is unavailable or unsuitable for the present duty — Criticality reflects the consequence of loss and the actual ability to recover. A standby that cannot carry the duty removes the recovery; price, a health rating and a clean record do not change the consequence.
2. A vendor offers to raise a work order automatically for every alert its model generates on a fictional operator's pumps. What do you approve?
A. Automatic work orders for every alert
B. Automatic work orders for high-confidence alerts, permits arranged later
C. No link to the maintenance system at all
D. A link in which each accepted warning gets an owner and disposition, with work still passing through the authorised process
D. A link in which each accepted warning gets an owner and disposition, with work still passing through the authorised process — Link the warning but preserve the authorised work process. Mass automatic tasks bury critical work, a confidence threshold supplies no scope, materials or permits, and refusing any link wastes the warning.
3. In the book's illustrative monitoring case, what happens to the NPV when avoided failures fall from four a year to two?
A. It falls from about $48,174 to about negative $95,936
B. It halves to about $24,087
C. It stays positive
D. It falls to exactly zero
A. It falls from about $48,174 to about negative $95,936 — Gross benefit halves to $60,000 but the $25,000 recurring cost does not, so net benefit falls to $35,000 and present value to about $84,064 against $180,000 up front. Halving the NPV forgets that fixed costs do not shrink.
Today’s Assignment
Write Section 16 of your plan: for one equipment or maintenance decision inside your Session 1 decision, the failure consequence, the actionable condition, the closed response chain and an investment case with a downside.
Choose one equipment or maintenance decision in your company inside your Session 1 decision. State the consequence of losing the equipment and your actual ability to recover, including the standby.
Say whether your records separate defects, functional failures, protective trips and planned interventions, and what a model trained on them would learn.
Name the actionable condition, its lead time, the role that defines the diagnostic question, and the cost of a false alarm and of a missed warning.
Trace the response chain from detection to confirmation, name the disposition owner by role, estimate stage times and identify the limiting step.
Write the investment case with a base and a downside, assumptions and limits beside the result. With no model in use, make the case for the simplest credible method, or none.
Say honestly what your business gets wrong today.
YOUR 90-DAY PLAN
Section 16 — Failure consequence, the actionable condition and the closed response chain.
How this is marked — the rubric
States the failure consequence with the real recovery position, including the standby
20
Defines the actionable condition with lead time, diagnostic owner by role and the cost of both errors
25
Traces the response chain stage by stage, names the disposition owner and identifies the limiting step
30
Shows base and downside cases with the assumptions and limits beside the result
25
Total · pass mark 70
100
Submitting session 16 Write your answer in Word or Google Docs. The first line must read Session: 16, 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.
Part Three · Processing, Reliability and Capital DeliverySESSION 17
Asset Integrity, Inspection and Remaining Life
Session 17 of 30 · 25-Minute Module
Why This Matters
A field can meet its production target while its protection against a major failure weakens, and nothing on the production report will tell you. This session is about the evidence that would: an asset you can identify across its records, a model asked a question it can bound, an action register that is honest about what is still open, and a review that protects assurance from commercial persuasion. AI can read more inspection records than any team. It cannot confer engineering acceptance on an uncertain asset. A leader who forgets that risks being reassured by a clean dashboard that sits on top of an unresolved finding.
25-Minute Module
0:00 – 6:00
Production is not integrity; start with an identifiable asset
THE DOCTRINE
An asset can produce steadily while its protection against a major failure weakens. Production performance cannot stand in for integrity assurance.
Integrity must make visible the condition of equipment, the quality of the evidence and the status of required action. AI can examine large inspection records; it cannot confer engineering acceptance.
Keep controlled information on what was designed, installed, modified and inspected: identity, material, service, limits and alteration history agreeing across engineering and maintenance records. Session 4's record discipline, applied to steel.
Thickness records are of limited value if locations cannot be matched over time. An improving trend may reflect a new measurement position, method or instrument. Retain uncertainty and method, trace anomalies to the original record, and preserve the reason and approving authority for any correction.
The official description of ISO 55001:2024 identifies requirements for an asset-management system (ISO, ISO 55001:2024, official standard description). A generic management framework does not replace the specialised engineering assessment a particular component needs.
When a corrosion trend improves and nobody did any work, ask what changed in the measurement before you celebrate the steel.
6:00 – 11:00
Ask the model a bounded question
Define the purpose. Analytics can prioritise records for review, find inconsistent observations, identify recurring degradation patterns or compare inspection schedules. Each question has an edge. Fitness for continued service is not one to hand to a model.
Remaining life depends on assumptions about degradation, loading, environment, inspection uncertainty and future use. It should show a range and the conditions that govern it. A date on a dashboard creates false certainty when those conditions are hidden.
Evaluate against relevant mechanisms and known cases, including sparse data. An average error across a broad population may conceal poor performance on the components with the greatest consequence.
THE DOCTRINE
Where an analytical finding affects operation, a competent integrity authority should review the evidence and the applicable engineering basis. The application may assist the assessment. It should not sign it.
Ask any remaining-life figure two questions: what is the range, and what must stay true for the range to hold.
11:00 – 17:00
Keep the action register honest
An integrity action needs five elements. Read them as questions for any entry in front of you.
Element
The question to ask of it
Finding
Clearly stated, and traceable to the original record and asset location
Consequence
What exposure remains while the action is open
Recommended response
Stated clearly enough that completion can be confirmed against it
Accountable owner
Named, so that someone can explain what prevents resolution
Required date
Reflects the technical basis and practical plan; any interim arrangement explicitly approved
Do not close an action because a work order exists; confirm the work was completed and its result accepted. A temporary measure is not permanent closure unless the authorised assessment supports it.
Ask to see overdue high-consequence actions and repeated extensions. A red status alone is insufficient: the review should say what prevents resolution, what exposure remains and what decision is required.
AI can connect findings across reports and histories, and reveal repeated wording that masks unresolved work. Investigate the substance; do not reward the appearance of a cleaner register.
A register that only ever gets greener deserves the hardest question in the meeting.
17:00 – 22:00
Changes in service, and assurance versus persuasion
A new duty, changed feed, revised operating pattern or extended life can alter the integrity basis. Management of change should examine the effects before implementation.
AT THE ASSET · THE WORKED CASE
Asset. The book's fictional pipeline has run within a familiar envelope for years. A commercial proposal changes the fluid and delivery pattern. Historical reliability under the old service does not prove fitness for the new one; the review needs the proposed conditions, materials, degradation mechanisms and any additional examination.
IN THE OPERATING REVIEW · THE PARALLEL
Review. Commercial should see the resulting limits and costs before promising delivery. For life extension, ask what evidence supports continued use, which assumptions must remain true and what future work is required, and put those obligations in the forecast.
Integrity professionals need a route to raise concerns against a clear commercial preference. Independent review matters most when continued operation or deferred expenditure rests on uncertain evidence.
THE DOCTRINE
A summary that removes caveats to improve readability can change the meaning of the assessment. The underlying qualifications must remain visible.
Measure evidence quality, action closure, repeated deferrals and the response to changed conditions. A falling backlog is useful only when it reflects necessary work completed and accepted.
The leadership obligation is to sustain a credible basis for operation. Analytics should strengthen that basis and expose its limits.
22:00 – 25:00
Before you close the laptop
Write one line: the asset in your Session 1 decision whose integrity basis you would least like to defend today, and why.
This session produces Section 17 of your 90-Day Intelligent Operating Plan: the bounded integrity question and an honest action register.
No process plant? Apply this to the equipment or facility you operate for clients, or the operator asset your service depends on.
Ask for the open high-consequence actions, oldest first. The order tells you more than the colour.
Key Terms
Technical authority
A person or function with the defined competence and responsibility to accept the relevant technical basis. The authority signs; the model does not.
Bounded question
A defined analytical task a model may perform, such as prioritising records for review or finding inconsistent observations, as distinct from an engineering acceptance it may not give.
Action closure
Confirmation that the required integrity work was completed and its result accepted. A raised work order or a temporary measure is not closure.
Change in service
A new duty, changed feed, revised operating pattern or extended life that can alter the integrity basis and must be assessed through management of change before implementation.
Self-Check — Before You Submit
1. After a new inspection campaign, wall-thickness readings on a fictional line improve although no repair was done. What is the right first reading?
A. The corrosion programme is working
B. It may reflect a new measurement position, method or instrument; trace the readings to the original records
C. Retrain the model on the new readings
D. Extend the inspection interval
B. It may reflect a new measurement position, method or instrument; trace the readings to the original records — An improving trend may reflect a change in measurement rather than condition. Crediting the programme, retraining or extending the interval all accept the trend before its basis is checked.
2. Which remaining-life output should an executive accept as decision support?
A. A single date, easiest to plan around
B. A good average error across the whole population
C. A range with governing conditions, reviewed by a competent integrity authority
D. A fitness-for-service conclusion signed by the application
C. A range with governing conditions, reviewed by a competent integrity authority — Remaining life should be a range with its conditions visible, reviewed by a competent authority. A single date gives false certainty, an average can hide the high-consequence components, and the application should not sign.
3. Your integrity report shows several high-consequence actions closed. Each was closed when a work order was raised; two rest on temporary measures. What do you do?
A. Reopen them until work is confirmed complete and accepted; treat temporary measures as open unless the authorised assessment supports closure
B. Accept the closures, since the work is planned
C. Accept them but show them in red
D. Have the AI tool reword the entries for clarity
A. Reopen them until work is confirmed complete and accepted; treat temporary measures as open unless the authorised assessment supports closure — A work order is not closure and a temporary measure is not permanent closure without the authorised assessment. A red status is insufficient, and rewording rewards the appearance of a cleaner register.
Today’s Assignment
Write Section 17 of your plan: for one asset in your business, the bounded integrity question a model may answer, an honest view of its action register, and how assurance is kept separate from persuasion.
Choose one asset inside or behind your Session 1 decision. Say whether its identity, material, service, limits and alteration history agree across your company's engineering and maintenance records.
Write the bounded question a model may answer about it, the question it may not, and the integrity authority by role. If a remaining-life figure exists, say whether it shows a range and its conditions.
Test up to five real register entries, described rather than copied if confidential, for finding, consequence, response, owner and date. Say honestly whether closed means accepted or only a work order raised.
Describe how an integrity professional in your company can raise a concern against a clear commercial preference, and assess one change in service or life extension in view.
Say what your company currently gets wrong. With no model in use, work from the integrity decision in front of you.
YOUR 90-DAY PLAN
Section 17 — The bounded integrity question and an honest action register.
How this is marked — the rubric
Tests asset identity and record agreement and names the gaps
20
Defines a bounded question, states what the model may not decide and names the integrity authority by role
25
Tests real register entries honestly for closure, overdue high-consequence actions and repeated extension
30
Describes a real route for raising concerns and assesses a change in service before implementation
25
Total · pass mark 70
100
Submitting session 17 Write your answer in Word or Google Docs. The first line must read Session: 17, 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
Integrity beyond productionIdentifiable asset recordsBounded model questionsHonest action registersAssurance over persuasion
Part Three · Processing, Reliability and Capital DeliverySESSION 18
Power, Utilities and Site Resilience
Session 18 of 30 · 25-Minute Module
Why This Matters
Instrument air does not sound like a leadership subject. Yet an operating plan can fail because a utility that looked secondary becomes the controlling constraint. Power, instrument air, water, fuel, cooling and communications support the plant's ability both to operate and to recover. This session asks you to map which dependencies can stop several services at once, to compare energy choices on the service they actually deliver, and to plan for the uncomfortable middle state in which something is degraded but still sending plausible numbers. The AI service you are planning belongs in that map too. If it fails, your team must still know what to do.
25-Minute Module
0:00 – 5:00
The secondary service that controls the plan
Utility reliability deserves the same system-level attention as production equipment. In Session 3's terms, the binding constraint for a given period may sit in a generator, the instrument-air supply or water treatment rather than in the wells.
THE DOCTRINE
Installed capacity is different from dependable capacity. State what is available in the present condition and what can be sustained during maintenance or an upset.
Dependable capacity includes what controls availability beyond the machine itself: fuel supply, spare parts and operator competence.
The executive question is narrow and useful: which dependencies can stop multiple services at once.
Ask for dependable capacity. Installed capacity is what the nameplate says.
5:00 – 11:00
Map the dependencies
Build a practical view of the site's utility sources, consumers and essential dependencies. Look for apparently independent services that share a common supply, control component or support arrangement. Two services that share a supply were never fully independent.
Utility
What the book asks you to establish
Power and fuel
Dependable capacity now and during maintenance or upset, with fuel supply, spares and competence included
Water
Sources, uses, storage and disposal; quality and treatment capacity; balance exceptions tracked with hydrocarbon-record care
Instrument air and cooling
Consequence-based attention, because a modest capacity shortfall can affect several process units
Communications and the AI service
Whether essential work continues, current procedures are accessible locally and stale data can be identified when the service is unavailable
Utility maintenance should reflect the actual impact of failure rather than a low equipment purchase price. A water volume trend without the required quality evidence may be commercially or operationally incomplete.
AI can combine operating histories and examine load, trips and fuel use; the first benefit may be a clearer understanding of the system. Keep it anchored in engineering: a correlation between two trips is not proof of a shared cause. It is a reason to examine the records and physical connections.
The cheapest equipment on site can still stop three units. Price is not consequence.
11:00 – 16:00
Compare energy choices on a consistent basis
THE DOCTRINE
An energy decision should compare delivered service at the required reliability. Capital cost alone can favour a scheme that is expensive to operate or unable to meet the site's duty.
A consistent basis includes fuel, maintenance, relevant losses, planned outages, backup arrangements and the costs of interruption. State the time horizon and any assumptions about tariffs or supply. Where a price is volatile, use a range and retain the sensitivity.
For self-generation, grid supply or a hybrid, show when each source is available and which loads it can support. An average annual energy balance can hide a short period in which essential loads cannot be met.
An optimiser should respect the approved operating and protection arrangements. It may recommend a dispatch pattern, but it should not alter protective functions without the separate engineering and governance process. On the Autonomy Ladder, that is recommendation, not control.
When two energy options are compared, ask whether both were costed at the same reliability. If not, the comparison is not finished.
16:00 – 22:00
Plan for degraded operation
Resilience requires a defined response when part of the utility system is unavailable. The site's operating procedures should establish priorities, permissible conditions and recovery arrangements through competent engineering and operational review.
The AI service belongs in that assessment. The book asks whether essential work can continue without communications or analytics, whether current procedures are accessible locally, and whether the team can identify stale data.
THE DOCTRINE
A completely unavailable service is often easier to recognise than a degraded one that continues to send plausible information. Exercise partial failure, not only total loss.
A simulation can rehearse a loss of power or communication and test authority, coordination and recovery. Do not assume simulated timing matches real equipment response without validation.
AT THE ASSET · THE WORKED CASE
Asset. The book's teaching case: a team proposes a more efficient generation schedule. The model reduces fuel consumption by assuming all generating sets can be switched freely. Maintenance reports that one set has a temporary restriction; operations reports that a transition requires a particular load arrangement.
The correct response is to revise the feasible set and obtain the required operating review.
IN THE OPERATING REVIEW · THE PARALLEL
Review. The earlier optimum was not executable. It can remain a technical exploration, but it should not be reported as a delivered saving.
After an approved trial, compare fuel use and service reliability under comparable conditions, accounting for load and other relevant differences. A quieter operating period should not be credited entirely to the optimisation.
If the plan only works when every set is available, it is a plan for a site you do not have.
22:00 – 25:00
Before you close the laptop
Write one line: the utility whose loss would stop your Session 1 decision soonest, and whether its capacity is installed or dependable.
This session produces Section 18 of your 90-Day Intelligent Operating Plan: your utility dependencies, a consistent comparison and the degraded-operation plan.
If you run no site, map the utilities your base, workshop or office depends on, or the operator's utilities your service relies on at the client site.
Pop Quiz 4 — Reliability, Integrity and Resilience opens now: five questions, closed book, covering Sessions 16 to 18.
The site leader owns resilience across services. Leadership funds the recovery arrangements the plan depends on.
Key Terms
Dependable capacity
What a utility can actually deliver in its present condition and sustain during maintenance or an upset, including the effect of fuel, spares and competence. Distinct from installed capacity.
Common dependency
A shared supply, control component or support arrangement that can stop several apparently independent services at once.
Consistent basis
Comparing energy options on delivered service at the required reliability, including fuel, maintenance, losses, outages, backup and interruption costs over a stated horizon.
Degraded operation
Operating within an authorised fallback when part of a utility or information service is unavailable or unreliable, with defined priorities, permissible conditions and recovery.
Self-Check — Before You Submit
1. A hybrid power proposal shows annual energy supply comfortably above annual demand. What can that view hide?
A. The capital cost of the scheme
B. A short period in which essential loads cannot be met
C. The fuel price used in the base case
D. The number of generating sets installed
B. A short period in which essential loads cannot be met — An average annual energy balance can hide a short period in which essential loads cannot be met; the review must show when each source is available and which loads it supports. Capital cost, fuel price and set count are visible elsewhere and do not answer the reliability question.
2. Your generation optimiser reports a fuel saving. Maintenance then tells you one set has a temporary restriction the model ignored. What do you direct?
A. Report the saving, since the model is validated
B. Remove the restriction from the maintenance plan
C. Let the optimiser adjust protective settings
D. Revise the feasible set, obtain the operating review and treat the result as technical exploration, not a delivered saving
D. Revise the feasible set, obtain the operating review and treat the result as technical exploration, not a delivered saving — The earlier optimum was not executable. Reporting it claims value never delivered, overriding maintenance moves the risk, and an optimiser must not alter protective functions outside the separate engineering and governance process.
3. Why does the book insist that resilience exercises include partial failure?
A. A degraded service still sending plausible information is harder to recognise than one completely unavailable
B. Total failures cannot be simulated
C. Partial failures always cost more
D. Partial-failure timing is already validated
A. A degraded service still sending plausible information is harder to recognise than one completely unavailable — A dead service is obvious; a degraded one sending plausible data is not, so teams must practise spotting doubtful inputs. Total loss can be simulated, the book makes no cost claim, and simulated timing needs validation.
Today’s Assignment
Write Section 18 of your plan: the power and utility dependencies behind your Session 1 decision, one energy or resilience choice compared on a consistent basis, and the degraded-operation plan.
Map the utilities your decision depends on in your company: power, fuel, instrument air, water, cooling and communications, as relevant. For each, state installed and dependable capacity and what controls availability.
Mark every place where apparently independent services share a common supply, control component or support arrangement.
Compare one energy or resilience choice in front of your business on delivered service at the required reliability: fuel, maintenance, losses, planned outages, backup and interruption costs, with the horizon stated and ranges for volatile prices.
Write the degraded-operation plan: priorities, permissible conditions, recovery arrangements and decision authority by role. Include any AI or analytics service: what continues without it, where procedures sit locally and how stale data are identified.
Name one partial-failure case your company should exercise.
Say honestly what your company currently gets wrong about utilities. Service companies work from their own base or the client-site utilities their work depends on.
YOUR 90-DAY PLAN
Section 18 — Our utility dependencies, a consistent comparison and the degraded-operation plan.
How this is marked — the rubric
Maps dependencies with dependable, not installed, capacity and identifies common dependencies
30
Compares one choice on delivered service at the required reliability, with horizon, assumptions and ranges
25
Writes a degraded-operation plan with priorities, authority and recovery, including the AI service
30
Names a partial-failure case to exercise and gives an honest account of current weaknesses
15
Total · pass mark 70
100
Submitting session 18 Write your answer in Word or Google Docs. The first line must read Session: 18, 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 4 — Reliability, Integrity and Resilience 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.
Part Three · Processing, Reliability and Capital DeliverySESSION 19
Engineering Projects and Commissioning Discipline
Session 19 of 30 · 25-Minute Module
Why This Matters
A project can finish construction and still fail the operator. The book's test of completion is stricter: the operator can use the asset safely, maintain it, obtain the intended service and accept the records and obligations that come with it. This session follows a project from the operating problem it should remove, through design with operations in the room, a schedule that is physically possible and commissioning to agreed criteria, to a handover that is protected rather than assumed. It ends with the book's case of a handover that looks nearly complete, a clear test of what a completion percentage can tell you and what it cannot.
25-Minute Module
0:00 – 5:00
Begin with the operating problem
THE DOCTRINE
The project is not complete when construction is complete. It is complete when the operator can use the asset safely, maintain it, obtain the intended service and accept the relevant records and obligations.
The sponsor should state the constraint the project is intended to remove and the evidence for it. Additional capacity has little value if another part of the system remains limiting: Session 3's binding constraint, now with a capital budget.
Compare alternatives before fixing scope: repair, revised operating practice, commercial coordination or a smaller investment. An expansion may still be right, but it should earn that choice through comparison.
Connect technical capability to saleable output, timing and cash. Approvals, feed, route access, utilities, staff and working capital belong in the project plan, not in a footnote after sanction.
Ask the sponsor which constraint disappears on the day the project starts up, and which one takes its place.
5:00 – 10:00
Bring operations into design
Operations and maintenance should help define access, maintainability, equipment selection, isolation, instrumentation, training and documents while design choices remain open.
IOGP's discussion of early operations integration describes how input from future users during projects supports operability, maintainability and safer outcomes (IOGP, Building safety in: how early operations integration in projects prevents future risks). The book's practical recommendation is to assign that input to named people and decision gates rather than rely on informal consultation.
AI can find inconsistent equipment identifiers, missing deliverables or contradictory requirements across a large document set. Each finding should lead to a traceable question or change; the engineering authority retains responsibility for the accepted design.
Retrieval tools must show the document version they used. A plausible answer from a superseded drawing can be worse than no answer, and access to a draft does not make it approved.
If your maintenance lead first sees the layout at handover, the access problems are already built in.
10:00 – 15:00
Make the schedule physically possible
A project schedule should represent dependencies, access, approvals, procurement, fabrication, installation, testing and handover. Activity durations should have an evidence basis appropriate to the decision.
An algorithm can explore alternatives and show the effect of a late item, but it must expose its assumptions about resources and parallel work. As in Session 16, impossible overlaps are not acceleration.
THE DOCTRINE
Distinguish contingency for uncertainty from funds for known omitted scope. If a required item is known, include it. Hiding it inside contingency makes later reporting less honest.
Forecast completion as a range where uncertainty is material. A date chosen for reporting convenience should not displace the planning basis.
Contingency is for what you do not yet know. What you already know belongs in the scope.
15:00 – 21:00
Commission to agreed criteria and protect the handover
Commissioning needs a defined plan, authority and acceptance criteria connecting installed equipment, completed tests, documentation, training and readiness for service. AI can organise punch items and identify missing evidence. It should not declare a system ready because a percentage of tasks is closed.
AT THE ASSET · THE WORKED CASE
Asset. The book's fictional project reports 98% of punch items closed. Remaining: a minor paint repair, an incomplete equipment record and an unresolved item affecting readiness for service. The percentage weights them equally.
The readiness authority classifies them by significance and acceptance basis. An AI summary may organise the list but must keep each item's status, source and qualification, and must not infer acceptance from a similar item closed elsewhere.
IN THE OPERATING REVIEW · THE PARALLEL
Review. One consequential item can govern the next stage. Leadership sees the reason for the restriction and the work to remove it; the sponsor resources the controlling work and communicates any effect on commercial commitments.
A temporary arrangement needs the applicable competent review, conditions and ownership.
The acceptance record states what was verified, who accepted it, which limitations remain and how they will be resolved. Temporary arrangements need ownership, not an assumption that operations will absorb them.
Handover needs usable drawings, equipment information, spares, maintenance requirements, procedures, competence and support, budgeted by the project. Asset identifiers should transfer cleanly into maintenance systems; commissioning data should keep operating conditions and distinguish tests from normal production.
Party
What it owns
Sponsor
The investment outcome
Engineering
The accepted design basis
Operations
Readiness and sustained use
AI
No acceptance; it helps connect their evidence
Leadership
Making those responsibilities meet before the asset changes hands
After start-up, compare realised constraint relief, costs, output and reliability with the sanction case. Capacity without value points to the commercial and operating dependencies.
A completion percentage is progress information. Acceptance is a decision someone must take and record.
21:00 – 25:00
Before you close the laptop
Write one line: the project or change in your company that touches your Session 1 decision, and the operating constraint it is meant to remove.
This session produces Section 19 of your 90-Day Intelligent Operating Plan: the project from the operating problem, its commissioning criteria and a protected handover.
If your company delivers projects rather than receiving them, write from your side of the handover. If no project is in view, use the change in front of you.
The handover should be a meeting of responsibilities, not a transfer of unresolved problems.
Key Terms
Early operations integration
Input from the future users of an asset during project design, assigned to named people and decision gates, to support operability, maintainability and safer outcomes.
Contingency
Funds held for uncertainty. Distinct from known omitted scope, which belongs in the estimate and should not be hidden inside contingency.
Acceptance record
The commissioning record of what was verified, who accepted it, which limitations remain and how they will be resolved.
Data handover
The clean transfer of asset identifiers and commissioning data into maintenance and operating systems, with tests distinguished from normal production.
Self-Check — Before You Submit
1. A fictional project reports 98% of punch items closed. One open item affects readiness for service. The contractor asks to proceed. What is the right next step?
A. Proceed, since 98% is high enough
B. Have the readiness authority classify the remaining items by significance, since one consequential item can govern the next stage
C. Let the AI tool close items similar to ones accepted elsewhere
D. Close the paint repair first to raise the percentage
B. Have the readiness authority classify the remaining items by significance, since one consequential item can govern the next stage — A percentage weights every item equally when the engineering assessment does not. The readiness authority decides significance. AI must not infer acceptance from a similar item, and closing easy items improves the number, not readiness.
2. A design-review assistant answers a maintainability question confidently but cannot show which document version it drew on. What is the principal risk?
A. None, if the answer is plausible
B. Only that checking takes longer
C. A larger model is needed
D. An answer from a superseded drawing can be worse than no answer, so the version must be shown
D. An answer from a superseded drawing can be worse than no answer, so the version must be shown — Retrieval tools must show the version used, because a plausible answer from a superseded drawing can be worse than none. Plausibility is the danger, and a larger model does not fix missing provenance.
3. Your project team knows a required item is missing from the estimate. The project manager proposes covering it from contingency to avoid reopening the budget. What do you direct?
A. Include the known item in scope and keep contingency for uncertainty
B. Accept, since contingency exists for items like this
C. Accept, but quietly raise contingency by the same amount
D. Defer the item until after handover
A. Include the known item in scope and keep contingency for uncertainty — Contingency is for uncertainty; a known required item belongs in scope. Hiding it makes later reporting less honest, and deferring it pushes the problem onto operations at handover.
Today’s Assignment
Write Section 19 of your plan: one project or change in your company framed from the operating problem, with operations brought into design, commissioning criteria set and the handover protected.
Name the project or change in your company, the operating constraint it should remove and the evidence. Say what will remain limiting afterwards.
List the alternatives compared, or that should have been, and connect the chosen option to saleable output, timing and cash.
Name, by role, the operations and maintenance people who will shape the design, and the decision gates that record their input.
Separate known omitted scope from contingency and give completion as a range.
Set the commissioning acceptance criteria, describe the acceptance record, and list the handover deliverables, data handover and post-service review.
Say honestly what your company gets wrong on projects. Service companies write from their side of the handover.
YOUR 90-DAY PLAN
Section 19 — The project from the operating problem, its commissioning criteria and a protected handover.
How this is marked — the rubric
Frames the project from a named operating constraint, compares alternatives and states the cash dependencies
25
Assigns operations input to named roles and decision gates, not informal consultation
20
Separates known scope from contingency and forecasts completion as a range
15
Sets commissioning criteria and an acceptance record that weight items by consequence, not count
25
Defines handover deliverables, data handover and the post-service review
15
Total · pass mark 70
100
Submitting session 19 Write your answer in Word or Google Docs. The first line must read Session: 19, 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
Problem-led project framingOperations in designHonest contingencyCriteria-based commissioningProtected handover
Part Four · Protecting Assets and Operating RelationshipsSESSION 20
Evacuation, Pipelines, Terminals and Custody Transfer
Session 20 of 30 · 25-Minute Module
Why This Matters
A barrel at the wellhead is not yet a delivered barrel. It may need treatment, storage, transport, measurement and acceptance before anyone can invoice it, and the route can determine both how much you produce and how much you are paid. This session opens Part Four of the book. It asks you to describe the route as it actually operates, to reconcile the custody evidence rather than force every system to agree, to compare routes by what they deliver rather than what they charge, and to measure what reaches the customer. If your company moves no hydrocarbons, the same chain still runs between your work and your cash.
25-Minute Module
0:00 – 5:00
Define the available route
The evacuation review connects the asset's operating plan with route capacity, reliability, custody-transfer evidence, contractual entitlement and settlement. Those responsibilities cross several departments and sometimes several companies.
Describe the route as it operates today: capacity, operating periods, storage, maintenance restrictions, access arrangements and dependencies on third parties. Distinguish contracted capacity from capacity that is actually available, as Session 18 distinguished installed from dependable power.
THE DOCTRINE
A line on a map is not a ready contingency. An alternative route may require preparation, approval, equipment or different product handling.
AI can compare route scenarios and outage effects on storage and production, if physical and contractual constraints are included. A scenario that exceeds tank capacity or assumes unapproved access is an investigation case, not an executable schedule.
Ask what the alternative route would need before it could carry your barrels next week.
5:00 – 11:00
Reconcile custody evidence
Maintain consistent records of dispatched, received, measured, allocated and invoiced quantities. This carries Session 4's record and Session 12's volume definition to the boundary where custody changes hands.
An observed imbalance can arise from timing, measurement uncertainty, conditions of measurement, inventory movement, allocation rules or an actual loss. Investigate the plausible causes with the relevant evidence before anyone names a culprit.
THE DOCTRINE
Explain the basis of differences rather than forcing every system to display the same number. An anomaly detector cannot by itself establish theft or liability.
The Nigerian Upstream Petroleum Measurement Regulations 2023 provide a primary regulatory reference for applicable upstream measurement arrangements (NUPRC, Nigerian Upstream Petroleum Measurement Regulations, 2023). The operator's legal and measurement specialists should confirm the requirements relevant to each location and approval.
AI can retrieve records and flag inconsistent units or missing tickets. The custody-transfer authority approves corrections and preserves the original record. Commercial and finance should be able to trace each invoice to the accepted quantity basis.
When every system shows the same number, ask who made them agree, and where the original went.
11:00 – 17:00
Compare routes through delivered contribution
The lowest quoted transport fee may not produce the highest value. Include preparation, fixed charges, variable charges, handling, credible losses, interruption and payment timing within the selected decision boundary. The book's invented one-year route proposal shows why, with assumed inputs.
Line
Working
Result
Incremental saleable volume
400 barrels per day × 120 operating days × 95% saleability
45,600 barrels
Contribution after the $2 route variable charge
45,600 × ($25 − $2)
$1,048,800
Fixed availability payments
$30,000 × 12 months, although the window is 120 days
$360,000
Annual net benefit
$1,048,800 − $360,000
$688,800
Present value at 12%, benefit at year-end
$688,800 ÷ 1.12
$615,000
NPV after $250,000 preparation paid immediately
$615,000 − $250,000
$365,000
Look at the fixed charge: twelve months of payments to earn a 120-day window. A comparison on the variable fee alone would miss $360,000, about a third of the $1,048,800 contribution after the variable charge.
The limits travel with the result: one year, pre-tax, no financing or working-capital effects. The $25 must exclude route costs already deducted; the 45,600 barrels must be truly incremental and not credited to another project; monthly timing should replace the year-end simplification before approval.
Ask for delivered contribution, not the tariff. The tariff is one line of it.
17:00 – 22:00
Design the operating agreement and measure what reaches the customer
Route performance depends on coordination: nomination, scheduling, communication, measurement, maintenance notification, dispute handling and authorised interruption, agreed through the applicable agreements and competent review. On a shared route, do not assume you can improve your schedule by displacing another party's allocation.
AT THE ASSET · THE WORKED CASE
Asset. On a fictional shared evacuation route, scheduling with another user fails repeatedly, and the asset team absorbs each miss by adjusting storage and production.
Each workaround is competent. None removes the cause.
IN THE OPERATING REVIEW · THE PARALLEL
Review. Leadership should know who can resolve a cross-company issue and how quickly. A repeated scheduling failure may require a commercial decision rather than another operational workaround.
When the route forecast changes, record the evidence and the resulting production, storage and cash consequences, and change the integrated plan with it.
Keep gross physical movement separate from net entitlement and invoiced value. A route forecast needs a documented baseline and an owner.
AI helps the team anticipate constraints and reconcile evidence. Leadership creates the agreements, resources and authority that allow the hydrocarbons to move and their value to be collected.
If the same scheduling failure appears in three weekly reviews, it is no longer only an operations problem.
22:00 – 25:00
Before you close the laptop
Write one line: the route by which your product or service reaches the customer, and the point at which you can no longer prove the quantity.
This session produces Section 20 of your 90-Day Intelligent Operating Plan: the available route, reconciled custody evidence and what reaches the customer.
If your company does not run this activity, a non-operated partner applies it to the operator's route and its own entitlement; a service company traces its work from completion through client acceptance to invoice and cash; others use the counterpart they depend on or their nearest equivalent decision.
Practice Exam 4 — Midpoint opens now: fifteen questions covering Sessions 1 to 20, the halfway check on the doctrine so far.
A barrel is delivered when it is measured and accepted, and its value counts when it is collected, not when it leaves the wellhead.
Key Terms
Available capacity
The route capacity actually usable today, after operating periods, storage, maintenance restrictions, access and third-party dependencies. Distinct from contracted capacity.
Custody evidence
Consistent records of dispatched, received, measured, allocated and invoiced quantities, with the basis of each difference explained and the original record preserved.
Delivered contribution
The value a route produces after preparation, fixed and variable charges, handling, credible losses, interruption and payment timing, within a stated decision boundary.
Net entitlement
The quantity a party is entitled to after allocation, kept separate from gross physical movement and from invoiced value.
Self-Check — Before You Submit
1. An anomaly detector flags a growing gap between dispatched and received quantities on a fictional route. A manager wants to notify the route operator of theft. What do you direct first?
A. Notify the route operator of suspected theft immediately
B. Investigate the plausible causes, from timing and measurement to allocation rules and actual loss, with the relevant evidence
C. Adjust the received figures to match dispatched so the systems agree
D. Retrain the detector to reduce the alarm
B. Investigate the plausible causes, from timing and measurement to allocation rules and actual loss, with the relevant evidence — An imbalance has several plausible causes, and an anomaly detector cannot by itself establish theft or liability. Forcing the systems to agree destroys the evidence the custody-transfer authority must preserve, and retraining the detector silences the question without answering it.
2. In the book's illustrative route case, why is the fixed cost $360,000 when the additional production window is only 120 days?
A. Preparation costs are spread across the year
B. The 12% discount rate is applied monthly
C. The saleability factor reduces the volume
D. Fixed availability payments of $30,000 a month are due for all 12 months
D. Fixed availability payments of $30,000 a month are due for all 12 months — The case assumes fixed availability payments are due for all twelve months. Preparation is a separate $250,000, discounting is applied once at year-end, and saleability affects volume, not fixed cost.
3. Which three quantities does the book say a route review must keep separate?
A. Gross physical movement, net entitlement and invoiced value
B. Fees, claims and cash receipts
C. Route availability, losses and inventory exposure
D. Contracted capacity, tank capacity and pipeline length
A. Gross physical movement, net entitlement and invoiced value — The book keeps gross physical movement separate from net entitlement and invoiced value because each answers a different question. The other groups are measures or capacities, not the separation the rule describes.
Today’s Assignment
Write Section 20 of your plan: the available route for your company's product or service, the reconciled custody evidence and a measure of what reaches the customer.
Describe the route by which your company's product or service reaches the customer as it operates today: capacity, operating periods, storage, restrictions, access and third-party dependencies. Distinguish contracted from available capacity, and say what the alternative would actually require.
Trace one recent period's quantities through dispatched, received, measured, allocated and invoiced. Explain each difference by its basis, and name, by role, who approves corrections and where the original record is kept.
Compare two routes or arrangements, or the current route and one change, on delivered contribution, with assumptions and limits stated. A simple worked comparison is enough.
Name the measures you would review monthly to show what reaches the customer, keeping gross physical movement, net entitlement and invoiced value separate, and name the owner of the route forecast.
If your company does not run this activity: a non-operated partner works from the operator's route and its own entitlement; a service company from completed work to client acceptance, invoice and cash; others from the counterpart they depend on or their nearest equivalent decision.
Say honestly where your business currently loses sight of quantity or value.
YOUR 90-DAY PLAN
Section 20 — The available route, reconciled custody evidence and what reaches the customer.
How this is marked — the rubric
Describes the route as it operates today and distinguishes contracted from available capacity
20
Traces custody quantities, explains each difference by its basis and names the correction authority by role
30
Compares routes on delivered contribution with assumptions and limits stated
25
Defines customer measures that separate physical, entitlement and invoiced value, with a named forecast owner
25
Total · pass mark 70
100
Submitting session 20 Write your answer in Word or Google Docs. The first line must read Session: 20, 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 4 — Midpoint 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.