30-Session Programme · 25 Minutes Per Day · Free · For leaders in Nigeria’s oil and gas sector
WEEK 3 OF 6 · WELLS, PRODUCTION AND PROCESSING · SESSIONS 11–15 · WORK AT YOUR OWN PACE
Part Two · Subsurface, Wells and ProductionSESSION 11
Well Integrity, Workovers and Intervention Economics
Why This Matters
Session 10 took a well to delivery. Today someone proposes to work on a well that is already producing, and the book opens with a distinction that is easy to blur: the duty to maintain integrity is not the choice to pursue an optional production improvement. Both require evidence; they are not ranked on the same basis. The chapter then shows, with simple assumed numbers, a workover whose successful case looks attractive and whose probability-weighted case loses money. I want you to leave able to repeat that arithmetic in an investment committee, and to ask what would change the probability rather than asking a model for a more optimistic forecast.
25-Minute Module
0:00 – 5:00
An integrity duty is not an optional improvement
- Integrity work may be necessary to control an unacceptable condition. A discretionary workover should compete for capital through its expected operating and financial value.
- THE DOCTRINE
A profitable forecast cannot cancel an integrity obligation, and an integrity obligation should not be disguised as a production-growth case.
- What the application can help establish: well history, trends, tests and previous interventions can point to candidate causes of underperformance, and pattern analysis helps direct attention across a large portfolio. The diagnosis still needs engineering review, because similar symptoms may arise from different conditions. The model should show the evidence it used and where observations are missing.
- A candidate intervention should state what the work will change, the probability and duration of a useful response, the risk of failure, the effect on integrity and the resources needed. A ranked list without those elements is not ready for approval. Keep three other options open: gather further evidence, choose another intervention, or decline.
- No wells? Apply today to the operator intervention you support or fund, or your nearest equivalent: a repair whose benefit is uncertain and whose cost is certain.
When a proposal arrives, ask first whether it is a duty or a choice. The answer decides which queue it joins.
5:00 – 12:00
The book's workover calculation, recomputed
- The book labels this example plainly: assumed inputs for teaching, not an actual asset forecast, service quotation or Nigerian market price. Inputs: initial cost USD800,000; if technically successful, 100 barrels per day in Year 1 (the book does not state the days, but its figures require a full 365-day year); 90% availability; 95% saleability; an assumed contribution of USD25 per saleable barrel; 20% annual decline; three years at a 12% annual discount rate.
Year Saleable barrels Contribution Discounted at 12% 1 100 × 365 × 0.90 × 0.95 = 31,207.5 USD780,187.50 ÷ 1.12 = USD696,595.98 2 31,207.5 × 0.80 = 24,966 USD624,150.00 ÷ 1.2544 = USD497,568.56 3 24,966 × 0.80 = 19,972.8 USD499,320.00 ÷ 1.404928 = USD355,406.11 Total USD1,549,570.65 - Successful-case NPV: USD1,549,570.65 less USD800,000 is USD749,570.65. Alone, it looks like an easy approval.
- Expected NPV: assume a 40% probability of that successful response and no incremental production if it fails. The cost is incurred in either outcome. So 0.40 × USD1,549,570.65 is USD619,828.26, less USD800,000, which is negative USD180,171.74.
- Break-even probability: USD800,000 ÷ USD1,549,570.65 is approximately 51.63%, before any additional costs or omitted consequences. Below that, the work destroys value on these assumptions.
- THE DOCTRINE
A positive successful-case NPV does not make the proposal attractive on the probability-weighted basis. The value of 51.63% belongs to the stated assumptions; recalculate it when they change.
- These are deliberately bounded screening calculations. They exclude corporate income tax, financing, working capital, terminal value and incremental closure costs. An actual sanction case must include the relevant asset and contractual basis.
Write the probability on the cover of the paper. A successful case without one is half a decision.
12:00 – 19:00
What evidence would change the choice, and who approves it
- Investigate whether the success probability can be improved, or the scope and cost changed. That is more useful than asking the model for a more optimistic forecast. Better evidence can also strengthen the decision to decline: a more confident estimate need not be a higher one.
- As in Session 8, a diagnostic test has value only if its possible results affect the decision, after its cost and delay. If it cannot distinguish the important failure modes, it adds information without decision value. If the uncertainty is an irreducible condition or a delivery route outside the intervention's scope, another model may add little.
- Test the other assumptions. A restored well cannot earn its contribution while a downstream constraint prevents delivery, and that constraint must not be deducted twice under different labels. A delay may move cash flows later or cut production within a fixed horizon; applying both without a separate reason overstates the penalty.
- AT THE ASSET · THE WORKED CASE
- Asset. In a fictional onshore operator, the wells team has a candidate with two plausible causes of underperformance. One targeted investigation could tell them apart; another model run could not.
IN THE OPERATING REVIEW · THE PARALLEL- Review. The investment committee sees the 40% case and asks what the investigation costs, how long it delays the work and which result would change the decision, including declining. Same loop, seen from the money.
- Approval is split by competence. The wells authority owns the technical and integrity basis; the asset manager owns the portfolio choice; finance verifies assumptions, timing and funding; procurement and logistics confirm that execution is feasible.
- The approved case retains its ranges and stop conditions. If the diagnosis changes, recalculate and record why; do not keep the earlier headline return on a quietly changed basis. Afterwards, compare actual cost, downtime, incremental saleable output and persistence with the approved case. A short initial response is not proof of a three-year result.
Ask your wells team why success is uncertain before you ask them how likely it is.
19:00 – 25:00
Before you close the laptop
- THE DOCTRINE
The purpose of AI in this decision is to improve diagnosis, candidate comparison and learning. The authority to accept the technical risk and commit capital remains with the operator.
- Write down one intervention on your Session 1 thread, whether it is a duty or a choice, its successful case, the probability you would honestly assign and the evidence that would move it.
- That is Section 11 of your plan: the intervention case, its probability and the evidence that would change it.
A well-ranked opportunity earns capital through its own case, not through a desire to show that the analytical programme produces approvals.
Key Terms
Successful-case NPV
The discounted value of an intervention if it works, less its cost. It says nothing about how likely success is.
Expected value
A probability-weighted result, distinct from any single realised outcome.
Break-even success probability
The success probability at which expected value is zero. It belongs to the stated assumptions; recalculate it when they change.
Value of information
The expected improvement in a decision from obtaining additional evidence, compared with its cost, delay and limitations.
Self-Check — Before You Submit
1. A workover paper shows a successful-case NPV of USD749,570.65 and an assumed 40% success probability. The sponsor asks the analytics team to rerun the model for a stronger forecast. What do you decide?
- A. Approve it, because the successful-case NPV is positive
- B. Rerun the model until the forecast improves
- C. Ask why success is uncertain and whether evidence, scope or cost could change the case, then recalculate break-even
- D. Decline all workovers until the model reaches 51.63% accuracy
C. Ask why success is uncertain and whether evidence, scope or cost could change the case, then recalculate break-even — On the book's assumptions the expected NPV is negative USD180,171.74, so the work is to change probability, scope or cost, not the forecast's optimism. The positive successful case is the trap; 51.63% is a break-even probability, not an accuracy target.
2. A well has an integrity condition that must be controlled, but the remediation shows a weak production return. How does the book treat it?
- A. The obligation stands and is not ranked as a growth case
- B. It competes for capital with discretionary workovers on expected NPV
- C. It is deferred until the production case improves
- D. It is presented as a production-growth case so that it ranks higher
A. The obligation stands and is not ranked as a growth case — The book separates integrity duty from optional improvement. Ranking it with discretionary workovers is the tempting error; disguising it as a growth case is the one the book names.
3. A diagnostic test is proposed before a workover. When does it have value?
- A. Whenever it adds information about the well
- B. When it is cheaper than the workover
- C. When it produces a more confident estimate
- D. When its possible results could change the decision, after its cost and delay are counted
D. When its possible results could change the decision, after its cost and delay are counted — A test has value only if its results affect the decision. One that cannot distinguish the important failure modes adds information without decision value, however cheap or confident.
Today’s Assignment
Write Section 11 of your plan: the intervention case for one decision in your company, with its successful case, probability, downside and the evidence that would change the choice.
- Choose one intervention on your Session 1 thread. If your company runs no wells, use the operator intervention you support, a partner workover you are asked to fund, or your nearest equivalent. State first whether it is an integrity duty or an optional improvement.
- Write the successful case in the book's shape, with your figures labelled as estimates. If you have none, say which you need and who holds them.
- State the probability of success and its evidence, then the expected value and break-even probability, with cost counted in either outcome. Name the downside: failure case, integrity effect, downtime and any downstream constraint, counted once.
- Name one investigation that could change the probability, scope or cost, its cost and delay, and the result that would make you decline. Name owners by role, stop conditions and the post-intervention comparison.
- The honest part: where does your company currently approve on the successful case alone, or rank interventions without a stated probability? Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 11 — The intervention case, its probability and the evidence that would change it.
How this is marked — the rubric
| Separates integrity duty from optional improvement and states the successful case with explicit, labelled assumptions | 25 |
| States the probability with its evidence and computes expected value and break-even correctly, with cost counted in both outcomes | 30 |
| Names one investigation whose possible results would change the decision, with its cost, delay and the result that would mean declining | 25 |
| Names owners by role for technical basis, portfolio choice, finance and execution, with stop conditions and a post-intervention comparison | 20 |
| Total · pass mark 70 | 100 |
Submitting session 11
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: 11, 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 versus discretionary workProbability-weighted screeningBreak-even reasoningValue of informationPost-intervention verification
Part Two · Subsurface, Wells and ProductionSESSION 12
Production Optimisation and Deferment Accountability
Why This Matters
A daily production report can explain yesterday without helping the team improve tomorrow. It may list every loss accurately while the same causes recur. The book's remedy: connect each loss to a diagnosis, an owner, a feasible action and an outcome. Last session's workover only proves its value if the barrels it restores are measured honestly, and that depends on the volume definitions and the deferment ledger we build today. I would ask you to look hard at your own morning meeting. If several teams claim the same recovered barrels, your company is reporting value it achieved once.
25-Minute Module
0:00 – 6:00
Know which volume is being discussed
- Think of the chain from the wellhead to the buyer: each link has its own number, and each number answers a different question.
- THE DOCTRINE
Production capability, actual production, saleable output, entitlement and delivered volume are different measures. Say which one you mean, and how it relates to the others, before the discussion starts.
The figure What it must not be treated as A forecast of well capability A forecast of collected revenue A facility total A figure comparable with individual estimated well rates, without a reconciliation basis A net entitlement A volume to add to gross operated output as though they were independent A virtual flow estimate useful for picking a test candidate An accepted custody-transfer quantity - Virtual flow estimation estimates flow between physical observations. It can assist surveillance, but its uncertainty and validation matter. The review needs agreed definitions, aligned periods and source records: Session 4's data foundation made practical.
- If your company is a non-operated partner, your discipline starts with entitlement against gross. If it is a service company, work with the operator volume your service affects and the record you will be judged on.
When two figures disagree in a meeting, first ask whether they are the same measure.
6:00 – 12:00
A deferment ledger that can be acted on and reconciled
- Each material loss should have a cause category, evidence, owner, expected duration and recovery action. Where the cause is unresolved, say so rather than assigning a convenient label.
- Distinguish avoidable loss from an operating restriction that cannot presently be removed, and an equipment outage from the broader consequence it causes. Otherwise shared events are counted several times.
- Rank actions by consequence, feasibility and resources. A small recurring loss can deserve attention if it can be resolved cheaply and reliably; a larger theoretical recovery may deserve delay if the execution conditions are absent. AI can recognise recurring patterns, but an exception that remains in an analytical application has not reached the operating team. The recommendation must connect to work management.
- The book's teaching example: a well interruption, a compressor outage and export unavailability overlap on the same day. Each team calculates the barrels affected by its own event. Added together, the three can exceed the entire day's feasible sales loss.
- The record should identify the period, physical constraint, feasible baseline, interacting events and actual output. Agree a method for attributing the system loss without claiming the same barrels several times, and keep the event-level detail even when one event takes the principal allocation. Engineers need every failure; finance needs a reconciled quantity basis. Neither is served by deleting inconvenient overlapping events.
- An AI tool can align event times and propose a reconciliation for review. The asset manager approves the basis with production and finance.
Ask to see the ledger for the worst day of last month, and add up the claims.
12:00 – 19:00
Optimise within an approved envelope, decide in the morning meeting
- An optimisation application should state the constraints it uses: safe limits, equipment capacity, fluid behaviour, measurement uncertainty and product commitments. A proposed change needs technical review proportionate to its consequence, and the application should show why it recommends the change and the conditions under which that reasoning holds.
- A short-term increase is not a sustainable improvement. A change can raise output while increasing wear, water handling, energy use or later intervention need. A bounded trial needs a defined comparison, duration and stop criteria, and a record of other operating changes, so that the result is not attributed to the model without evidence.
- The morning meeting should concentrate on exceptions that need a choice. Which loss is growing? Which diagnosis is uncertain? Which accepted action is waiting for approval or a part? Which constraint has changed?
- THE DOCTRINE
The report can be assembled automatically from controlled sources. The decision cannot be delegated to the assembly process.
- AT THE ASSET · THE WORKED CASE
- Asset. At a fictional operator, one well's estimated rate drops overnight. It may be the well, or the estimate drifting from its last physical test.
IN THE OPERATING REVIEW · THE PARALLEL- Review. The morning meeting splits it into a reconciliation and a field check, each with an owner and a review date, instead of one undifferentiated red list.
- Record decisions in the meeting with the responsible owner, the review date and the supporting evidence. If the team rejects a recommendation, record why.
A data discrepancy and an operating problem both need resolving. They rarely need the same person.
19:00 – 25:00
Verify the recovered value, then write
- Measure an improvement through saleable and deliverable output under the agreed basis. Explain how baseline decline, planned activity and changing availability were treated.
- Where maintenance and production applications contribute to the same recovery, reconcile attribution. Commercial and finance should confirm the economic consequence, because additional output can change cost, inventory, delivery and receipt timing. The benefits ledger should show those relationships.
- THE DOCTRINE
The company gains one set of restored barrels, not a separate set for every team involved. Use the reconciled ledger when assessing an intervention's outcome.
- The intelligent production review, the book says, is a management arrangement supported by analysis: it recognises problems earlier, assigns practical responses and tests whether they changed the result.
- That is Section 12 of your plan: the volume definition, our deferment ledger and how recovered value is verified.
If every project in the portfolio claims success, add their barrels and compare the total with the sales.
Key Terms
Deferment
Production or delivery not achieved relative to an explicitly defined feasible basis. Its causes and overlapping events require reconciliation.
Benefit attribution
The method used to determine how much of an observed change belongs to a particular intervention, so that one recovery is not claimed several times.
Operating envelope
The approved conditions and limits within which an asset or activity may operate, established through the applicable competent process.
Virtual flow estimation
A calculated estimate of flow between physical observations. It can assist surveillance but must be validated and used only for the purpose the operator approved.
Self-Check — Before You Submit
1. A well interruption, a compressor outage and export unavailability overlap on one day. Each team reports its own loss, and the total exceeds the day's feasible sales loss. What do you do?
- A. Accept the largest single figure and delete the other two events
- B. Approve, with production and finance, a method that claims each barrel once and keeps event-level records
- C. Report all three figures, since each team calculated its own event correctly
- D. Ask the AI tool to choose the principal cause and publish its figure as final
B. Approve, with production and finance, a method that claims each barrel once and keeps event-level records — The book requires a reconciled system loss with event-level detail retained. Deleting events fails the engineers; adding the three figures fails finance. AI may propose, but the asset manager approves the basis.
2. Finance proposes to use a virtual flow estimate, validated for choosing well-test candidates, as the accepted custody-transfer quantity. What is the book's position?
- A. Acceptable if the estimate is recent
- B. Acceptable if the model is machine learning rather than physics-based
- C. Required, because estimates are more frequent than physical tests
- D. Not automatically: an estimate useful for one approved purpose is not thereby an accepted custody-transfer quantity
D. Not automatically: an estimate useful for one approved purpose is not thereby an accepted custody-transfer quantity — The estimate should be used for the purpose the operator approved. Frequency or recency is the tempting argument, but neither changes the approved use or the uncertainty.
3. Your morning meeting shows one long red list that mixes data discrepancies with operating problems. What is the first improvement?
- A. Separate them, because one may need a reconciliation and the other needs work
- B. Remove the data discrepancies from the report
- C. Automate the meeting so the report assembles the decisions
- D. Add more signals so the list is complete
A. Separate them, because one may need a reconciliation and the other needs work — Both need resolution, but one needs a reconciliation and the other work. Automating the decision is the tempting error: it cannot be delegated to the assembly process.
Today’s Assignment
Write Section 12 of your plan: the volume definition your company's decision depends on, one actionable deferment-ledger entry, and how recovered value will be verified.
- Start from your Session 1 decision. A non-operated partner works from entitlement against gross; a service company uses the operator volume its service affects; a business with no production uses its nearest equivalent loss of output.
- Define the volume: which of the five measures, how it relates to the others, source record, period and owner. Say where your company currently mixes them.
- Draft one ledger entry for a recent material loss: cause category, evidence, owner, duration, recovery action, avoidable or restriction, interacting events. If the cause is unresolved, say so. Test one recent day when events overlapped for double counting.
- State how recovered value will be verified: baseline, treatment of decline, planned activity and availability, attribution between teams, and who in commercial and finance confirms it.
- The honest part: what does your company's morning meeting currently report that nobody acts on? Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 12 — The volume definition, our deferment ledger and how recovered value is verified.
How this is marked — the rubric
| Defines the volume precisely, relates it to the other measures and names where the company currently mixes them | 25 |
| Drafts a ledger entry with cause, evidence, owner by role, duration and recovery action, and states any unresolved cause honestly | 30 |
| Tests one overlapping event for double counting and proposes an attribution method that keeps event-level detail | 25 |
| States the verification basis, including baseline treatment and the commercial and finance confirmation | 20 |
| Total · pass mark 70 | 100 |
Submitting session 12
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: 12, 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
Volume definitionsDeferment accountabilityLoss reconciliationBounded optimisation trialsBenefit verification
Part Two · Subsurface, Wells and ProductionSESSION 13
Flow Stations, Processing and Flow Assurance
Why This Matters
A facility can run within every individual equipment limit while the overall process performs poorly. Separation is unstable, chemicals are consumed without a clear response, water handling becomes limiting and product quality varies. The team manages each symptom separately. Last session's ledger told you where barrels went missing; this session asks you to read the plant as one set of relationships rather than a list of alarms. It is uncomfortable, because it questions routine spend and shows how a throughput dashboard can report success while the hydrocarbons that pay the bills barely move.
25-Minute Module
0:00 – 5:00
Connect the physical picture
- The operating record should describe the current configuration, fluid properties, equipment condition, control settings and laboratory results, and record changes to them. Think of the plant drawing in the control room: it is useful only while it shows the plant as it now is.
- A model trained on stable historical conditions may be unreliable after a tie-in, an equipment modification or a material change in fluids. The process authority should decide when the application requires re-evaluation.
- Know which method you are using. Physics-based calculations and engineering judgement establish the feasible relationships; statistical methods help detect deviations. A digital twin is useful when it is maintained and supports a defined decision.
- THE DOCTRINE
A label such as digital twin should not imply a complete or continuously accurate representation. State what is included, what is simplified and what observations keep it current.
- No processing facility? Apply today to the operator decision you serve, the facility decision you approve as a partner, or your nearest equivalent plant, utility or quality control.
Ask when the plant last changed, and whether anyone re-evaluated the model afterwards.
5:00 – 11:00
Chemical treatment needs a testable purpose
A chemical programme should identify The question a leader can ask The condition it addresses What problem would return if we stopped dosing? The evidence of effectiveness What observation shows it is working? The dosing basis Why this rate, and under what conditions? The relevant limits Where does more become harmful or wasteful? - Consumption alone says little about performance. AI can assist the analysis of dose, conditions and response, but watch for confounding change: a favourable laboratory result may coincide with a change in feed or throughput rather than a dosing improvement.
- A proposed change needs competent review, a test within approved conditions and a look at consequences elsewhere. The book's teaching scenario shows improved separation alongside deteriorating discharged-water quality: the local gain was not worthless, but the choice must satisfy the whole set of constraints.
- Sand, scale, wax and corrosion each need mechanism-specific evidence. A common analytics platform does not make their diagnosis interchangeable, and a signal may justify investigation without establishing the mechanism. Some affect throughput; others affect integrity or safe operation. Repeated short-term treatment can hide an unresolved root cause.
When a chemical line rises in the budget, ask for the response it bought, not the litres it used.
11:00 – 16:00
Laboratories are part of operating control
- A laboratory result needs a sample basis, method, time and applicable specification. A value without those details may be difficult to interpret.
- Predicted quality can support scheduling or early investigation. It should not automatically replace the accepted release test. The product and environmental authorities determine the role of each method.
- THE DOCTRINE
When online and laboratory measurements disagree, reconcile the difference. Do not choose the value that makes the operating report look better.
- Reconciling means checking sample timing, calibration, methods and process conditions. Quality exceptions should then reach the people who can change the process, hold material or investigate the measurement. The path from result to action is part of the design.
A laboratory result that reaches no one who can act is a record, not a control.
16:00 – 21:00
Debottlenecking with a full consequence
- A proposed facility improvement should state the limiting capacity, the evidence that it is binding and what becomes limiting after the change. Session 3's binding constraint moves; the paper must say where.
- Include utilities, maintenance, storage, water handling, quality and evacuation. A throughput increase that cannot be delivered, or creates an unacceptable environmental consequence, is not a complete improvement. Show the operating window, change approvals and the team's readiness; a short performance test may not establish long-term reliability.
- AT THE ASSET · THE WORKED CASE
- Asset. The book's fictional facility: an upstream change raises total fluid throughput. Oil sales barely move, power use rises and water handling approaches its available capacity.
IN THE OPERATING REVIEW · THE PARALLEL- Review. A throughput dashboard reports an improvement because it has not separated the streams or tested the downstream effect. The executive paper should instead show saleable hydrocarbon output, water demand, energy, incremental cost and the next controlling constraint.
- Review oil, gas and water balances on consistent conditions. The higher fluid rate may be useful, but its value needs the actual hydrocarbon contribution and handling cost. The process authority confirms the constraints, the asset team the available well actions, and any fluid allocation respects the integrity and reservoir considerations of Sessions 9 and 11.
- The chapter's management tool is a facility constraint map linked to the operating decision record: where value is restricted, what evidence supports the diagnosis and which authorised action can improve the system.
Ask what becomes limiting next before you approve what is limiting now.
21:00 – 25:00
Before you close the laptop
- THE DOCTRINE
The operator needs a coherent operating choice, with the plant's limits visible before the change is approved.
- Write down one processing or flow-assurance decision on your Session 1 thread, one chemical or laboratory control and the purpose you could test it against, and one debottlenecking idea with what would become limiting next.
- That is Section 13 of your plan: the physical picture, testable treatment and the full consequence of change.
- Pop Quiz 3 opens now: five questions, closed book, covering Sessions 11 to 13.
The plant does not read the dashboard. Make sure the dashboard reads the plant.
Key Terms
Digital twin
A maintained representation of a facility used to support a defined decision. The label does not imply a complete or continuously accurate model; state what is included and simplified.
Confounding change
A second change, such as in feed or throughput, that coincides with the one being evaluated and could explain the observed result instead.
Release test
The accepted, authorised verification of product quality against the applicable specification. A quality prediction does not automatically replace it.
Facility constraint map
The chapter's management tool, linked to the decision record, showing where value is restricted, the evidence for the diagnosis and the authorised action that can improve the system.
Self-Check — Before You Submit
1. A chemical supplier reports that a higher dose improved separation. The same week, feed and throughput also changed. What do you ask for?
- A. A renewal at the higher dose, since the laboratory result improved
- B. A cut in dosing, because consumption rose
- C. Competent review of the confounding changes, a test within approved conditions and a check of consequences elsewhere
- D. An AI model to confirm the supplier's conclusion
C. Competent review of the confounding changes, a test within approved conditions and a check of consequences elsewhere — A favourable result may coincide with a change in feed or throughput. Accepting the improvement is the tempting error; a model alone cannot remove the confounding change.
2. The online analyser and the laboratory disagree on a quality value an hour before the operating report is issued. What is right?
- A. Report the value that is within specification
- B. Reconcile the difference by checking sample timing, calibration, methods and process conditions
- C. Report the online value, because it is more frequent
- D. Average the two and report the mean
B. Reconcile the difference by checking sample timing, calibration, methods and process conditions — The book says to reconcile and never to choose the value that makes the report look better. Frequency and averaging are tempting shortcuts that avoid the investigation the disagreement requires.
3. Which element must a debottlenecking proposal contain under the book's doctrine?
- A. A throughput gain measured during a short performance test
- B. A vendor's digital twin of the new configuration
- C. A forecast of total fluid handled
- D. The limiting capacity, evidence that it is binding, and what becomes limiting after the change
D. The limiting capacity, evidence that it is binding, and what becomes limiting after the change — The book asks for the binding limit, its evidence and the next constraint. A short test may not establish long-term reliability; total fluid handled is the chapter's own dashboard trap.
Today’s Assignment
Write Section 13 of your plan: the physical picture behind one processing or flow-assurance decision in your company, one chemical or laboratory control with a testable purpose, and the full consequence of one debottlenecking idea.
- Choose one decision on your Session 1 thread. A service company uses the service it supplies and the operator decision it feeds; a non-operated partner, the facility decision it approves; a business with no processing, its nearest equivalent plant, utility or quality control.
- Connect the physical picture: configuration, fluid properties, equipment condition, control settings and laboratory results. Name the last material change and whether any model was re-evaluated after it.
- Give one chemical or laboratory control a testable purpose: the condition addressed, the evidence of effectiveness, the dosing or sample basis and the limits. Say how you would rule out a confounding change.
- State the full consequence of one debottlenecking idea: the limiting capacity and the evidence it is binding, and what becomes limiting next across utilities, maintenance, storage, water handling, quality and evacuation, with the approvals required.
- Name, by role, who receives a quality exception and can act on it. The honest part: which consumption or throughput figure does your company currently treat as evidence of performance? Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 13 — The physical picture, testable treatment and the full consequence of change.
How this is marked — the rubric
| Connects the physical picture, names the last material change and states whether models were re-evaluated after it | 25 |
| Gives one chemical or laboratory control a testable purpose and a way to rule out confounding change | 25 |
| States the full consequence of the debottlenecking idea, including the next limiting constraint and the approvals required | 30 |
| Names by role who receives quality exceptions and can act, and identifies the activity measure currently mistaken for performance | 20 |
| Total · pass mark 70 | 100 |
Submitting session 13
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: 13, 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 3 — Interventions, Deferment and Flow 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 3 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Reading the plant as a systemTestable chemical treatmentLaboratory reconciliationDebottlenecking disciplineMethod selection
Part Three · Processing, Reliability and Capital DeliverySESSION 14
Gas Processing, Compression and Monetisation
Why This Matters
The book is precise about when gas becomes a business: when the operator can gather it, condition it to the required specification, deliver it through a dependable route and collect payment. A production forecast establishes only part of that chain. Today you hold the whole commitment in one review: available gas, treatment capacity, compression, power, customer demand, delivery obligations and credit exposure. This is where an attractive volume becomes a contract you must honour and a receivable you must collect, and where flare reduction becomes a project.
25-Minute Module
0:00 – 6:00
Establish the delivery basis
- A gas volume means little until you know the basis on which it is stated. Define the forecast consistently: measurement conditions, composition assumptions and treatment losses. Distinguish raw gas, processed gas, fuel use, reinjection, flare and sales. Volumes on different reference conditions should not be combined without an agreed conversion.
Function What it owns Engineering The conversions and their uncertainties Commercial The contractual delivery basis Finance Revenue forecasts built on the same saleable quantities - A dashboard cannot resolve inconsistent definitions by making numbers look precise. The gas balance should reconcile the principal streams. When a meter is unavailable, disclose the estimate and its basis, and keep observed, calculated and assumed quantities distinct.
- THE DOCTRINE
An exception remains a question for a competent reviewer. Automatically replacing an inconvenient measurement can conceal a developing equipment problem or a contractual dispute.
- No gas plant? Apply today to the operator's gas decision you serve, the delivery your company depends on as buyer or partner, or any product you promise to a specification and a date.
Ask which of the six streams a number belongs to before you ask whether it is good.
6:00 – 12:00
Optimise compression within the envelope
- Compression suits analytical support because production, power consumption, equipment condition and ambient conditions interact. The useful output is a feasible choice within the validated envelope. The model should retain constraints on availability, specification, operating limits and planned maintenance, and say which one controls. A lower calculated energy cost has little value if the state cannot be sustained safely.
- Compare it with the engineering basis and the simple rules experienced operators already use. Test uncommon but consequential conditions. If a critical input is missing or stale, the application should withdraw its recommendation or name a defined restricted use.
- The IEA describes opportunities for AI across energy operations, including production and maintenance (IEA, Energy and AI, 2025). That supports examining a local application; it does not establish a site's return or authorise a change to its control system.
- THE DOCTRINE
Start with advisory use. The shift team needs to understand the proposed change, the relevant limit and the way to revert.
- Any expansion into automatic action must follow the governance and technical assurance taught in Session 6.
If the shift team cannot explain how to undo a recommendation, it is not ready to follow one.
12:00 – 18:00
Price the obligation; treat flare reduction as a project
- Commercial value depends on the contract as well as the plant. Examine quantity commitments, quality, delivery points, interruption provisions, pricing, settlement and credit support with qualified commercial and legal reviewers. A forecast that sells every unit at the headline price can overstate the opportunity; include the costs to reach the customer and show scenarios where consequences are uncertain.
- AI can retrieve and compare clauses, but the reviewer checks the original clause, amendments and context; a summary never silently displaces the executed agreement.
- Customer credit is a separate decision. Reliable delivery to a customer that does not pay can increase the operator's funding requirement. Track overdue balances, concentration and the practical effectiveness of agreed security, and model cash receipts independently of recognised sales.
- Gas that is currently flared is not automatically a profitable gas project. A capture proposal needs a viable gathering arrangement, treatment concept, destination, operating plan and funding case, and it must distinguish the routine operating opportunity from safety-related requirements.
- The World Bank's Zero Routine Flaring by 2030 initiative establishes a public framework for addressing routine flaring (World Bank, Zero Routine Flaring by 2030 initiative text). A corporate commitment should become asset projects and measured progress. Ask which constraint prevents recovery and which investment would remove it.
- THE DOCTRINE
A gathering project without dependable offtake can leave the operator with additional equipment and the same unresolved commercial problem.
Delivered and invoiced is not the same as paid. Put both lines on the same page.
18:00 – 25:00
A gas-delivery review, then write
- AT THE ASSET · THE WORKED CASE
- Asset. The book's fictional plant can meet its monthly quantity target only by deferring a compressor service. Operations sees a delivery opportunity. Maintenance sees growing failure exposure.
IN THE OPERATING REVIEW · THE PARALLEL- Review. Commercial sees a customer obligation. Treasury sees an overdue account. The committee compares feasible alternatives: can another machine carry part of the duty, can the customer agree a revised schedule, what would an unplanned outage cost, and is further credit exposure authorised?
- No analytical score should combine those issues so completely that the committee cannot see them. Record the chosen delivery plan, maintenance action, commercial arrangement and cash assumption, assign an owner to each dependency, and review the actual result against those separate commitments.
- Write down the delivery basis for your decision, the obligation it carries and its price, and one improvement you would frame as a project with an owner, a constraint to remove and a funding case.
- That is Section 14 of your plan: the delivery basis, the priced obligation and the improvement framed as a project.
AI can help coordinate the evidence. Leadership must decide which obligations the operator can responsibly make.
Key Terms
Delivery basis
The stated measurement conditions, composition assumptions and treatment losses on which a gas quantity is forecast, reported and sold.
Gas balance
A reconciliation of the principal streams, raw, processed, fuel, reinjection, flare and sales, that explains material differences and discloses estimates.
Advisory application
A system that provides information or a recommendation for competent human review within its approved scope.
Credit exposure
The amount the operator stands to lose if a customer does not pay. Track overdue balances, concentration and the effectiveness of security, separately from sales.
Self-Check — Before You Submit
1. Your plant can meet this month's delivery target only by deferring a compressor service, and the customer has an overdue balance. An analytics tool returns one composite score recommending deferral. What do you do?
- A. Compare feasible alternatives with each issue visible, and record each commitment with an owner
- B. Follow the composite score, since it already weighs all four issues
- C. Defer the service, because delivery obligations come first
- D. Hold all deliveries until the overdue balance is paid
A. Compare feasible alternatives with each issue visible, and record each commitment with an owner — No score should combine the issues so completely that the committee cannot see them. Following it is the tempting shortcut; the other two skip the comparison of alternatives.
2. The IEA's Energy and AI report describes opportunities for AI in production and maintenance. How does the book use that source?
- A. As evidence of a typical return for compression optimisation
- B. As support for examining a local application only
- C. As regulatory approval for advisory use
- D. As a reason to move directly to automatic control
B. As support for examining a local application only — The source supports examining a local application; it does not establish a site's return or authorise a control change. Reading it as a return figure is the tempting misuse.
3. A team proposes capturing gas that is currently flared, citing the company's public flaring commitment. What does the proposal still need?
- A. Only the corporate commitment, since flared gas is free feed
- B. A forecast of flared volumes and the headline gas price
- C. Gathering, treatment, a destination with dependable offtake, an operating plan and a funding case
- D. An AI model of flare volumes
C. Gathering, treatment, a destination with dependable offtake, an operating plan and a funding case — Flared gas is not automatically a profitable project. Without dependable offtake the operator can be left with extra equipment and the same commercial problem.
Today’s Assignment
Write Section 14 of your plan: the delivery basis for one gas or delivery decision in your company, the price of the obligation it carries, and one flare-reduction or delivery improvement framed as a project.
- Choose one gas or delivery decision on your Session 1 thread. Without a gas plant, use the operator decision you serve, the delivery your company depends on, or your nearest equivalent promise to a specification and a date.
- Establish the delivery basis: the conditions and composition assumptions, the streams you distinguish, which quantities are observed, calculated or assumed, and who owns conversions, the contractual basis and the revenue quantities, by role.
- Price the obligation: quantity, quality, delivery point, interruption provisions, pricing and settlement, the costs to reach the customer, and the customer's credit. Show cash receipts separately from recognised sales, and use scenarios where a consequence is uncertain.
- Frame one flare-reduction or delivery improvement as a project: the constraint it removes, the gathering or delivery route, the destination and offtake, the operating plan, the funding case and an owner.
- If optimisation is involved, state its envelope and controlling constraint. The honest part: where does your company currently forecast revenue on a volume nobody has reconciled, or count delivery as if it were cash? Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 14 — The delivery basis, the priced obligation and the improvement framed as a project.
How this is marked — the rubric
| Establishes a consistent delivery basis that distinguishes streams and data types and names owners by role | 25 |
| Prices the obligation against the contract terms and treats customer credit and cash receipts separately from sales | 30 |
| Frames the improvement as a project with a removable constraint, dependable destination, operating plan, funding case and owner | 25 |
| States honestly where volume or delivery is currently mistaken for collectable value | 20 |
| Total · pass mark 70 | 100 |
Submitting session 14
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: 14, 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
Gas delivery basisEnvelope-bounded optimisationContract and credit disciplineFlare reduction as a projectCross-functional delivery decisions
Part Three · Processing, Reliability and Capital DeliverySESSION 15
Refining Yield, Quality and Product Value
Why This Matters
A refinery creates value by converting feed into products that meet specification and can be sold at an adequate contribution. Higher throughput helps only when the rest of that sentence stays true. For an integrated operator there is a quieter trap: crude sent to the refinery has an alternative value, and treating it as free makes processing look profitable by moving value from another part of the company. Today the book gives two short calculations: a positive campaign contribution, and the same campaign draining cash in the period. Any leader who approves on margin should be able to reproduce both.
25-Minute Module
0:00 – 5:00
Define the economic boundary
- First agree what the review concerns: a feed selection, a production campaign, a plant investment or the performance of the whole refinery. Each needs a different cost boundary.
Decision What the boundary includes Short-run campaign Incremental costs and available capacity Investment Capital, sustained operating costs, working capital, relevant taxes and eventual liabilities Corporate performance review Internal transfer assumptions reconciled with external group economics - Do not mix the cases. A campaign that contributes positively to cash does not prove the plant earns an adequate return on invested capital. A weak month does not alone establish that an asset should close.
- THE DOCTRINE
AI can compare feasible feed and operating scenarios, but the economic boundary must come from the decision owner. A mathematically elegant objective cannot correct an inappropriate definition of value.
- No refinery? Apply today to one product, quality or value decision: a crude or gas sale where quality meets price, the feed decision of a refinery you supply, or a service accepted on a client test.
Ask what the feed would have earned elsewhere before you ask what the refinery earned.
5:00 – 10:00
Connect prediction to laboratory evidence
- Quality prediction can flag an emerging problem before the next laboratory result. As in Session 13, the predicted property must stay distinguishable from the verified measurement.
- The quality authority defines how a prediction may be used. Screening, investigation and process adjustment are different permissions; as on the Autonomy Ladder, permission belongs to the approved use. Product release follows the applicable specification and the authorised verification process.
- A model can become unreliable when the input population changes, even if the software has not: feed composition, laboratory methods, equipment and process conditions all matter.
- Keep sample identity, collection time, analysis time and analytical method together. Otherwise a model may appear to predict quality accurately because the comparison pairs the wrong sample with the wrong operating period.
When a quality model looks perfect, check the sample timestamps first.
10:00 – 16:00
The book's campaign calculation, recomputed
- The book calls this an invented teaching case, not a forecast for a Nigerian refinery. A campaign processes 6,000 barrels of feed; all yields, prices and costs are assumptions.
Line Arithmetic Amount Diesel 6,000 × 70% = 4,200 bbl × $90 $378,000 Naphtha 6,000 × 20% = 1,200 bbl × $65 $78,000 Residue 6,000 × 5% = 300 bbl × $40 $12,000 Loss 300 bbl, no saleable product; saleable output 5,700 bbl $0 Revenue $378,000 + $78,000 + $12,000 $468,000 Feed opportunity cost 6,000 × $60 $360,000 Processing and logistics 6,000 × $6 = $36,000; 6,000 × $4 = $24,000 $60,000 Campaign contribution $468,000 − $420,000; ÷ 6,000 feed bbl = $8 $48,000 - The calculation excludes fixed plant costs, taxes, financing, capital recovery and working-capital changes. It is a campaign contribution illustration, not a whole-asset valuation.
- It assumes every product sells at the stated price. A lower realised price, an unsaleable batch, extra handling or a different yield changes the result; show the sensitivities rather than presenting $48,000 as dependable.
Put the feed opportunity cost on the first line, where nobody can lose it.
16:00 – 21:00
A profitable campaign can consume cash
- Now assume 20% of diesel revenue is still uncollected at the end of the review period: 20% × $378,000 is a $75,600 receivable. All other revenue is collected, and all $420,000 of campaign costs are paid in the period.
- Cash receipts are $468,000 − $75,600 = $392,400. Cash paid is $420,000. Net cash movement is negative $27,600, despite the positive $48,000 contribution. The gap between the two, $48,000 − (−$27,600) = $75,600, is exactly the outstanding receivable.
- THE DOCTRINE
A contribution margin should not be used as evidence that the campaign funds itself.
- The example ignores opening cash so the timing effect is visible. Treasury must examine the actual opening balance, committed facilities and payment calendar before execution. AI can show when cash is needed and which assumption supplies it; management decides whether the customer terms and funding are acceptable.
- AT THE ASSET · THE WORKED CASE
- Asset. At a fictional integrated operator, the planners' attractive campaign needs a product tank that is due for inspection and more laboratory capacity than the site has. On paper it pays; on the site it cannot move.
IN THE OPERATING REVIEW · THE PARALLEL- Review. The committee sees $48,000 of contribution; treasury sees negative period cash. A scenario that depends on unavailable capacity is labelled a proposed investment case, not an executable plan.
Ask treasury for the lowest cash day of the campaign, not the margin at its end.
21:00 – 25:00
Plan the entire campaign, then write
- A feasible schedule needs feed availability, suitable storage, available equipment, utilities, laboratory capacity, product tanks and dispatch arrangements, within maintenance and inspection restrictions. The refinery manager owns the operating plan; commercial owns the sales assumptions; finance verifies contribution and collection; the quality and integrity authorities keep their separate approvals; the COO resolves conflicts across the integrated system.
- Review margin, specification compliance, energy intensity, losses, inventory age, realised prices and receipts together: products customers accept, routes that move them and cash that reaches the bank.
- That is Section 15 of your plan: the economic boundary, evidence for quality and the cash test.
- Practice Exam 3 opens now: fifteen questions, closed book, covering Sessions 11 to 15. It closes Week 3.
Margin earns the approval. Cash decides whether the plan survives the month.
Key Terms
Feed opportunity cost
The value the feed could earn in its alternative destination. Treating internally supplied crude as free transfers value from another part of the company.
Campaign contribution
Campaign revenue less feed opportunity, processing and logistics costs. It is not a whole-asset valuation.
Economic boundary
The set of costs and values a decision includes, set by the decision owner and different for a campaign, an investment and a corporate performance review.
Working capital
Funds tied in operating balances such as inventory and receivables, net of applicable payables. Its timing can make a profitable plan consume cash.
Self-Check — Before You Submit
1. The campaign paper shows a $48,000 contribution and the sponsor says the campaign funds itself. 20% of diesel revenue will be collected after the period. What do you ask for before approval?
- A. Nothing further, since the contribution is positive
- B. A larger campaign to increase the contribution
- C. A lower feed opportunity cost so the margin is stronger
- D. A treasury check of receipts against payments, the opening balance, committed facilities and the payment calendar
D. A treasury check of receipts against payments, the opening balance, committed facilities and the payment calendar — On the book's assumptions the campaign moves cash by negative $27,600 because $75,600 is uncollected. A positive margin is the tempting evidence, and it is not proof of self-funding.
2. An integrated operator values the crude it sends to its own refinery at zero. What does the book say this does?
- A. Makes the processing business look profitable by transferring value from another part of the company
- B. Correctly removes an internal transfer from the analysis
- C. Is required for a short-run campaign decision
- D. Only matters for tax reporting
A. Makes the processing business look profitable by transferring value from another part of the company — The crude has an alternative destination or value, so the book requires an explicit feed opportunity cost. Treating it as a removable internal transfer flatters the refinery.
3. A quality prediction model performs well in testing. A planner proposes using it to release product without the laboratory test. What is right?
- A. Approve it, because the model has been tested
- B. Approve it for low-value products
- C. Keep release on authorised verification; the quality authority sets any narrower use of the prediction
- D. Replace the laboratory with an online analyser instead
C. Keep release on authorised verification; the quality authority sets any narrower use of the prediction — The book separates prediction from verified measurement and treats screening, investigation and adjustment as different permissions, with release following authorised verification. Good test performance is the tempting argument.
Today’s Assignment
Write Section 15 of your plan: the economic boundary for one product, quality or value decision in your company, how its prediction connects to verified evidence, and how a profitable plan could still consume cash.
- Choose one product, quality or value decision on your Session 1 thread. Without a refinery, use the nearest equivalent named in the session, or the campaign or offtake your company is asked to fund.
- Define the economic boundary: which decision type it is, which costs are in and out, the opportunity cost of the input and who sets it, and what a positive result would not prove.
- Connect prediction to evidence: the predicted property, the verified measurement it must match, who decides its permitted use, and how sample identity, time and method stay together.
- Run the cash test in the book's shape with your own figures, labelled as estimates: contribution, the amount outstanding at period end, cash paid and net cash movement. Name, by role, who checks the opening balance, facilities and payment calendar.
- Name one constraint that could stop the plan moving through the site, and say whether it is executable or a proposed investment case.
- The honest part: where does your company currently approve on margin and discover the cash position later? Aim for 400 to 900 words.
YOUR 90-DAY PLAN
Section 15 — The economic boundary, evidence for quality and the cash test.
How this is marked — the rubric
| Defines the economic boundary for the decision type, including the input's opportunity cost and what the result does not prove | 25 |
| Connects prediction to verified evidence with a named authority for permitted use and a sound sample record | 20 |
| Runs the cash test correctly with labelled estimates and names the treasury owner by role | 35 |
| Names the constraint that could stop the plan and classifies it honestly as executable or an investment case | 20 |
| Total · pass mark 70 | 100 |
Submitting session 15
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: 15, 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 3 — Wells, Production and Processing 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 3 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Economic boundary settingFeed opportunity costPrediction versus verificationContribution and cashWhole-campaign planning