THE INTELLIGENT EPC LEADERSHIP PROGRAMME
The contractor that can explain its delivery.
30-Session Programme · 25 Minutes Per Day · Free · For leaders in Nigeria’s engineering, procurement and construction sector
WEEK 4 OF 6 · CONSTRUCTION, COMMISSIONING AND ACCEPTANCE · SESSIONS 16–20 · WORK AT YOUR OWN PACE
Part Four · Construction, Commissioning and AcceptanceSESSION 16
Work Packaging and Construction Readiness
Why This Matters
Week 4 moves from supply to the workfront. A crew standing at a package it cannot execute costs your company twice: in idle hours and in the confidence of everyone told the area was ready. Chapter 16 asks a plain question before any technology: is this work executable today, by this crew, within accepted limits? You will learn why readiness is a conjunction rather than an average, why the look-ahead starts from the completion sequence, and how to design a bounded release service that reconciles evidence without ever releasing work itself. If you sit on the client, subcontractor or vendor side, apply it to the package your commitment feeds.
25-Minute Module
0:00 – 5:00
Start with work that can actually be executed
- The Construction Industry Institute describes Advanced Work Packaging as an integrated process spanning engineering, procurement and construction, with related work packages through delivery and turnover (Construction Industry Institute, Advanced Work Packaging Overview). The book's readiness model is an adaptation for executive decision-making, not a replacement for your project's accepted packaging method.
- A package is defined by scope, boundaries, predecessor work, released engineering, accepted material, access, permits, people, equipment, inspection, completion evidence and the commissioning system it serves, executable by the responsible crew within accepted limits.
- Then separate package definition, which says what the work is, from package release, which says the conditions are met now.
- THE DOCTRINE
A file containing the required forms is not evidence that the workfront is ready. Each condition needs a current state, owner and supporting record.
When someone tells you a package is ready, ask to see the record behind the one condition you would least like to be wrong about.
5:00 – 11:00
Readiness is a conjunction
- In plain terms, a conjunction passes only when every part passes, as a lift is not approved because most of its checks are signed. For a package with six mandatory conditions, readiness is true only when all six are met.
- THE DOCTRINE
Five met conditions and one unresolved isolation do not produce 83% permission to work. Averaging mandatory conditions can conceal the blocking one.
- Separate mandatory gates from productive preferences. A package can be authorised but less productive because a preferred tool is unavailable, if the approved method permits an alternative. It cannot proceed without a mandatory condition because its productivity forecast is favourable.
Condition state What it means for release Met Evidence accepted by a named owner, for a stated period Unmet Known open; blocks release if mandatory Unknown Stays visible; absence of a record is never converted to met Expired Was met; an enabling condition has lapsed or changed Not applicable Only with accepted justification - A binary ready flag is calculated only after these states are established. Readiness is time-dependent: a withdrawn drawing, damaged material or changed workfront can invalidate an accepted package, and the service should notify the responsible authority. AI can identify likely closure dates and draft a readiness brief; it must not mark a condition closed from an inferred relationship.
If your readiness report carries a percentage, ask what the percentage is permitted to authorise. The honest answer is nothing.
11:00 – 16:00
Look ahead from the completion sequence
- The sequence should support system readiness, not maximise disconnected installation volume. A large easy package can make progress look strong while a small dependency delays the milestone.
- Keep a look-ahead with package, planned start, missing condition, owner, closure action and latest useful decision date. When one technical input blocks several packages, show the shared constraint. If a condition stays open across several meetings, escalate the decision rather than repeat the description. With too few comparable cases, use the owner's stated forecast and show its uncertainty rather than fabricate a learned estimate.
- Track commitment reliability: what the team agreed to make ready against what was ready at the agreed time, with causes classified. Do not improve it by removing difficult packages from the denominator after the period closes.
- Put overdue management decisions beside overdue team actions. A coaching note to the foreman does not resolve a resource request awaiting approval.
A look-ahead that only describes the same constraint every week has become minutes, not management.
16:00 – 21:00
A bounded release service, and the walkthrough at week twelve
- Begin with one discipline and one area. The service matches accepted drawings, material status, predecessor completion, method, workfront conditions and authorisations by package, tag and revision. Its output is a reconciliation of conditions, not an electronic permit; manual release stays with the actual construction authority.
- Test it with withdrawn engineering, material received but not accepted, an expired condition, inconsistent identity and a predecessor marked complete without evidence, plus correctly ready packages. Measure false-ready recommendations separately from conservative holds, at the same cut-off. A hold because evidence is absent needs the owner; a hold because retrieval missed an accepted record needs the service repaired.
- Physical sequence stays outside free-form generation. A solver compares packages once eligibility, precedence, crew capacity and windows are represented; an unavailable authorisation excludes the activity rather than adding a penalty productivity can outweigh. In the book's hypothetical choice, W-02 shows more planned quantity but has an unresolved predecessor, so it stays ineligible and W-01 is selected.
- In the book's assumed walkthrough, at the proposed week-twelve start one supplier document is still pending review. The service identifies the gap; the authorised owner decides whether it blocks installation or a later completion transition. W-01 is released with its limits recorded, or held with a named closure action.
- AT THE WORKFRONT · THE WORKED CASE
- Workfront. A fictional EPC contractor's piping supervisor sees five of six conditions met and an isolation open. The service shows unmet, not 83%. The construction manager holds the package and moves the crew to executable work.
IN THE COMMITMENT · THE PARALLEL- Commitment. The project director, who owns cross-functional constraints, sees that the isolation depends on a client interface decision and escalates it to the client counterpart, not the foreman. The weekly outlook promises nothing that depends on it.
Better retrieval scores alone never justify automatic release; any expansion of authority needs its own assessment.
21:00 – 25:00
Carry accepted output to the next system stage
- After work, reconcile actual quantities, inspection evidence and installed configuration with the package as released; in the walkthrough, against revision B. A photograph may assist a visible-identity check but cannot establish hidden material, connection or test conditions.
- A finished package with an unresolved system interface stays in the next readiness review. Measure accepted work available to the crew at planned start, and the system stage it serves.
- Write down one package in your decision chain that was reported ready and was not. Name the condition that failed and who should have owned it.
- Today's assignment produces Section 16 — Executable work, readiness as a conjunction and a bounded release service.
Installation is not acceptance. Make sure your reporting never lets one stand in for the other.
Key Terms
Package release
The authorised decision, by the actual construction authority, that a defined package may be executed now within stated limits.
Readiness as a conjunction
Readiness is true only when every mandatory condition is met; an average of met conditions authorises nothing.
Latest useful closure date
The last date by which an open condition must close for the package to start as planned; the trigger for escalation.
False-ready recommendation
A service output proposing a package as ready when the accepted reference at the same cut-off shows a mandatory condition unmet.
Self-Check — Before You Submit
1. A package has six mandatory conditions. Five are met and one isolation is unresolved. What is its readiness?
- A. 83%, enough for a partial start
- B. Ready, if the productivity forecast is favourable
- C. Not ready; the open isolation blocks release
- D. Ready to mobilise while the isolation is chased
C. Not ready; the open isolation blocks release — Readiness is a conjunction: true only when all mandatory conditions are met. The 83% figure is the trap, because averaging conceals the blocking condition.
2. Your planner's optimiser penalises working without an authorisation but still selects the package because its output is high. What do you require?
- A. A larger penalty
- B. The authorisation as a hard constraint, so the activity is excluded rather than traded off
- C. Acceptance, since the solver has weighed the risk
- D. A sequence written by the language model instead
B. The authorisation as a hard constraint, so the activity is excluded rather than traded off — A penalty permits a trade-off the project does not permit. A larger penalty can still be outweighed; exclusion cannot.
3. The service held a package for want of an accepted material record that existed but retrieval missed. How is this classified?
- A. A service repair issue, distinct from a hold where evidence is genuinely absent
- B. A correct conservative hold
- C. A false-ready recommendation
- D. An owner's failure to close the condition
A. A service repair issue, distinct from a hold where evidence is genuinely absent — Absent evidence needs the owner to close a condition; a retrieval miss needs the service repaired. Calling it a correct hold hides the defect.
Today’s Assignment
Write Section 16 of your plan: executable work, readiness as a conjunction and a bounded release service for one package in your company's decision chain.
- Choose one package your decision chain depends on and define it as Appendix C does: scope boundary, system served, planned start, required finish, supervisor and each mandatory condition individually.
- State honestly how your company decides readiness today, and describe one recent case where a package reported ready was not.
- Classify each condition as mandatory gate or productive preference, with state, owner and latest useful closure date.
- Build the look-ahead from the completion sequence: which system the package serves and which small dependency could delay the milestone.
- Design a bounded release service for one discipline and one area: what it reconciles, its test cases, how false-ready results are measured and who releases. With no AI in use, design the manual reconciliation first; a simpler method may be the right answer.
- Clients, subcontractors and vendors: apply this to the package your commitment feeds.
YOUR 90-DAY PLAN
Section 16 — Executable work, readiness as a conjunction and a bounded release service.
How this is marked — the rubric
| Defines a real package with individually listed mandatory conditions, states and named owners by role | 30 |
| Treats readiness as a conjunction and separates mandatory gates from productive preferences | 25 |
| Builds the look-ahead from the completion sequence with latest useful closure dates and escalation | 20 |
| Specifies a bounded release service that reconciles evidence, keeps sequence outside free-form generation and leaves release with the construction authority | 25 |
| Total · pass mark 70 | 100 |
Submitting session 16
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: 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.Capabilities Covered
Executable package definitionReadiness as a conjunctionCompletion-led look-aheadBounded release designFalse-ready testing
Part Four · Construction, Commissioning and AcceptanceSESSION 17
Productivity, Resources and Subcontractor Delivery
Why This Matters
It is tempting to answer a low productivity figure with a request for more people or a complaint about the crew. Chapter 17 asks your company to look first at the work system: whether the work was ready, whether quantities were measured on one basis, and whether the hours being saved can actually be saved. You will work through the book's assumed productivity proposal and its selection of work under a resource constraint, both recomputed here. Then you will face two questions that test leadership more than analytics: how a subcontractor's reported progress becomes certified progress, and how workforce data can be used without becoming surveillance. Subcontractors and vendors will recognise this from the other side of the certificate.
25-Minute Module
0:00 – 5:00
Measure productive output on a consistent basis
- Productivity analysis needs accepted quantity, relevant hours, scope and conditions. Installed metres, tonnes or joints should not be mixed without a defined conversion or separate analysis. Hours on approved rework and hours spent waiting stay distinguishable.
- A model trained on reported progress can reproduce overstatement. Set the measurement rules and inspect the evidence before forecasting performance; where subcontractor reports are used, keep their acceptance and correction history.
- Compare like with like: design maturity, access, congestion, method, material availability and crew capability.
- THE DOCTRINE
A low output rate does not establish poor effort. The crew may have been assigned work that was not ready.
Before you discuss a crew's rate, ask how many of its hours were spent waiting for something the office owed it.
5:00 – 11:00
Use the analysis to change the work system
- The productive response may be better packaging, a different sequence, extra inspection capacity or corrected material supply. More labour can worsen congestion and supervision load; model the enabling constraints before assuming headcount scales output in a straight line.
- The book's assumed illustration: a crew uses 1,000 hours to install 200 accepted units. A proposal assumes four hours per unit on another 300 units. The figures are assumed, not labour-rate guidance.
Line Assumed calculation Result (recomputed) Current rate 1,000 hours ÷ 200 units 5 hours per unit Hours saved 300 units × (5 − 4) hours 300 hours Gross saving 300 hours × assumed ₦10,000 per hour ₦3 million Added cost ₦1.2 million preparation + ₦800,000 equipment ₦2 million Simplified net saving ₦3 million − ₦2 million ₦1 million - Now the uncomfortable part. If the hours are salaried and cannot be redeployed or removed, the immediate cash saving can be lower or zero. The value may be capacity or schedule, which needs separate evidence.
- THE DOCTRINE
Hours saved on paper are not automatically cash saved. If salaried hours cannot be redeployed or removed, the immediate cash saving can be lower or zero; capacity or schedule value needs separate evidence.
Ask your finance director which line of that table would survive contact with the payroll.
11:00 – 17:00
Case 3: selecting work under a resource constraint
- The casebook assumes three authorised packages competing for one specialist crew over a ten-day window: P needs 4 crew-days (teaching value 9), Q 6 (value 11), R 5 (value 10). The values are management preference units, not money or safety scores. Plainly: which combination fits ten days and is worth most? Maximise 9xP + 11xQ + 10xR with 4xP + 6xQ + 5xR ≤ 10, each x zero or one.
Selection Crew-days Value Feasible P and Q 4 + 6 = 10 9 + 11 = 20 Yes, the best P and R 4 + 5 = 9 9 + 10 = 19 Yes Q and R 6 + 5 = 11 11 + 10 = 21 No, exceeds 10 P, Q and R 15 30 No Q alone, R alone, P alone 6, 5, 4 11, 10, 9 Yes - A greedy ranking by value takes Q, then fails to fit R, and may or may not consider P depending on implementation. Enumeration solves this small case; a large portfolio may need a solver.
- Add a real dependency: R must finish to enable a system due in the window, so xR = 1. The best feasible plan becomes P and R, value 19. Management should not soften the system obligation into an ordinary preference without authority.
- Then P's released drawing is withdrawn, so xP = 0. With R mandatory, only R fits, using 5 of 10 crew-days. The model should report the 5 unused crew-days and the conflict. It should not fill them with unauthorised P.
Idle capacity honestly reported is a management signal; capacity filled with unauthorised work is a hidden risk.
17:00 – 21:00
Competence, certification and trust
- Resource models represent type, availability, competence, authorised work, calendars and location. Skills are not interchangeable merely because employees share a job title. If an assignment changes competence or supervision assumptions, route it for review; a low-cost schedule that depends on unqualified substitutions is not feasible.
- Define deliverables, measurement, documentation, quality conditions and payment basis in the subcontract, and link reported progress to accepted quantities. An AI summary of subcontractor activity should not become a certification without the responsible review. Track interface failures: a subcontractor can meet its local scope while another party's unfinished work prevents system acceptance.
- Avoid covert individual surveillance through a productivity project. Establish purpose, lawful processing and appropriate access; staff should know how observations are used and how errors are corrected.
- AT THE WORKFRONT · THE WORKED CASE
- Workfront. A fictional fabrication subcontractor's welding crew shows a low joint rate. The analysis finds the joints were released before the fit-up material was accepted. The response is corrected material supply, not a warning to the welders.
IN THE COMMITMENT · THE PARALLEL- Commitment. The main contractor's commercial lead receives the subcontractor's monthly claim. Progress is certified against accepted quantities and evidence; the AI summary is an input to the responsible reviewer, and the hours lost to the contractor's own late material are recorded as an interface failure, not as poor subcontractor performance.
A tool that shifts blame toward workers while ignoring missing drawings and material will weaken trust and deliver poor operational learning.
21:00 – 25:00
Write the section
- The management outcome is accepted output with dependable quality and usable resources. AI helps when it reveals constraints and supports feasible changes while preserving the dignity and accountability of the people doing the work.
- THE DOCTRINE
Measure the work system before making personnel conclusions. Represent competence, not job titles, and never present a schedule as feasible when it depends on unqualified substitutions.
- Write down one productivity figure your company reports today and the basis it is measured on. If you cannot state the basis, that is your finding.
- Today's assignment produces Section 17 — Productive output measured, the work system changed and competence preserved.
Before you pay for a productivity gain, ask whether the package was actually ready.
Key Terms
Accepted quantity
Installed output that has passed the agreed measurement and acceptance rules; the numerator the book requires for a productivity rate.
Mandatory selection
A package that must be chosen because it enables an obligation, such as a system due in the window; it is a constraint, not a preference.
Eligibility
Whether a package may be selected at all; a withdrawn drawing makes a package ineligible whatever its value.
Certification
The responsible review that turns reported subcontract progress into accepted quantity and payment basis.
Self-Check — Before You Submit
1. In the book's assumed illustration, what is the simplified net saving, and what can reduce the immediate cash saving to zero?
- A. ₦3 million; nothing, because the hours are saved
- B. ₦1 million; salaried hours that cannot be redeployed or removed
- C. ₦1 million; a higher hourly rate
- D. ₦2 million; the extra preparation cost
B. ₦1 million; salaried hours that cannot be redeployed or removed — 300 hours at an assumed ₦10,000 gives ₦3 million gross, less ₦2 million of preparation and equipment, leaving ₦1 million. ₦3 million ignores the costs, and salaried hours that cannot be redeployed or removed can save little or no immediate cash.
2. In Case 3, R becomes mandatory and P's drawing is withdrawn. What should the model report?
- A. Q and R, value 21
- B. P and R, value 19
- C. R alone, with 5 unused crew-days and the conflict reported
- D. R plus P to fill the capacity, flagged for review
C. R alone, with 5 unused crew-days and the conflict reported — With xR = 1 and xP = 0, Q and R need 11 crew-days and do not fit, so only R is feasible. Filling the remaining days with unauthorised P is exactly what the book forbids.
3. Your operations director wants camera and access-log data to rank individual welders for a productivity bonus. What is the right first response?
- A. Set purpose, lawful processing and access, tell staff how data is used and corrected, and measure the work system first
- B. Proceed quietly to avoid distorting behaviour
- C. Proceed, since higher output benefits the project
- D. Refuse all workforce analytics permanently
A. Set purpose, lawful processing and access, tell staff how data is used and corrected, and measure the work system first — The book rejects covert individual surveillance and asks for purpose, lawful processing, access and correction. Blanket refusal is the tempting overcorrection; the work system can still be analysed with trust.
Today’s Assignment
Write Section 17 of your plan: how your company measures productive output for one workfront in its decision chain, how it would change the work system, and how it protects competence and trust.
- Name one workfront in your decision chain and the productivity figure your company reports for it. State its quantity basis, hours basis and conditions, and honestly where the basis is inconsistent.
- Identify the enabling constraints behind a low rate (packaging, sequence, inspection capacity, material) before any headcount response.
- Price one improvement proposal with your own inputs labelled as assumptions, and state whether its value is cash, capacity or schedule.
- Set out one resource selection as Case 3 does: capacity, candidate packages, mandatory selections and eligibility. Say what the model must report when capacity is left unused.
- State how subcontract progress becomes certified progress in your business, and how workforce data is governed. Subcontractors and vendors: write it from your side of the certificate.
YOUR 90-DAY PLAN
Section 17 — Productive output measured, the work system changed and competence preserved.
How this is marked — the rubric
| States a consistent measurement basis for accepted quantity and hours, and admits where current reporting is inconsistent | 25 |
| Names enabling constraints and prices one proposal with labelled assumptions, separating cash from capacity or schedule value | 25 |
| Frames a resource selection with capacity, mandatory selections and eligibility, without filling capacity with unauthorised work | 25 |
| Links subcontract certification to accepted evidence and sets purpose, access and correction for workforce data | 25 |
| Total · pass mark 70 | 100 |
Submitting session 17
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: 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
Consistent productivity measurementWork-system diagnosisConstrained work selectionCompetence-preserving resourcingTrustworthy workforce analytics
Part Four · Construction, Commissioning and AcceptanceSESSION 18
Schedule Intelligence, Cost Control and the Estimate to Complete
Why This Matters
The schedule and the cost report are where your company's promises become numbers the board reads. Both can look precise while resting on defective logic or an inconsistent cut-off. Today covers two chapters, schedule first. You will see why a schedule is a model of delivery logic, why the critical path is set by the network rather than the size of the work, what a P80 date means, and why an expedite can recover less time than its sponsor claims. Then you will put cost and progress on one scope and forecast the remaining work rather than extrapolate the past. Every figure from the walkthrough and two casebook cases is recomputed.
25-Minute Module
0:00 – 6:00
The schedule is a model of delivery logic
- A dependable schedule represents the work, relationships, calendars, resources and actual status needed to explain delivery. The GAO Schedule Assessment Guide (GAO-16-89G, December 2015) is a public reference for schedule quality, critical-path analysis and schedule-risk assessment; it does not determine the contractual treatment of a Nigerian project's delay.
- Before adding AI, inspect missing logic, inappropriate constraints, excessive lags, unsupported progress and calendar inconsistencies. Keep baseline, current update and recovery scenarios separate, with the data date and cut-off.
- The critical path, plainly, is the chain of dependent work that sets the finish. Total float is how far other work can slip before joining it. Float is a schedule result, not an unconditional right to consume time; several parties may have an interest in it under the contract.
- The book's assumed walkthrough network, in weeks: engineering E 0 to 3; equipment P 3 to 10; foundations C 3 to 8; installation I waits for both, running 10 to 13; testing and handover T 13 to 16. E-P-I-T controls the sixteen-week finish; C has 10 − 8 = 2 weeks of float. That is not permission to start installation, and a brief saying foundations control because there are many civil tasks is wrong however plausible it sounds.
A package's financial value or physical size does not establish its schedule importance. Ask to see the logic.
6:00 – 11:00
Case 2: the recovery that recovers less
- The casebook's assumed network, in days: design A 10; procurement B 25 after A; civil C 20 after A; installation D 8 after B and C; testing E 5 after D. A runs 0 to 10, B 10 to 35, C 10 to 30, D 35 to 43, E 43 to 48. A-B-D-E controls; C has 5 days of float.
Assumed scenario B finish C finish Project finish Baseline 35 30 48 C delayed 3 days 35 33 48, within float C delayed 7 days 35 37 50 B delayed 4 days 39 30 52 B then expedited 3 days 36 30 49: 3 days recovered Same expedite, C at day 37 36 37 50: only 2 recovered - A procurement-only report could incorrectly claim three days of recovery; the integrated scenario shows two. A recovery case must show safety, quality, technical and funding constraints as well as time, with resource loading and cost matching the plan. The arithmetic shows network consequences, not contractual entitlement or site feasibility.
- THE DOCTRINE
A forecast changes the understanding of exposure. A funded, authorised and executed recovery changes the project outcome. The reporting system should preserve that distinction.
When someone presents a recovery, ask which other path becomes controlling once it succeeds.
11:00 – 15:00
Risk simulation and forecast quality
- Risk simulation runs the network many times with uncertain durations to see the spread of finish dates. Run the network for each scenario rather than buffering every activity and also adding a project contingency, which double counts. A three-point estimate is an elicitation input, not proof of a triangular distribution.
- A P80 date means 80% of simulated outcomes finish by that date under the represented assumptions. It is not an 80% guarantee of actual delivery.
- Evaluate learned forecasts on held-out periods or projects; a random split of overlapping updates leaks near-identical information. Measure error at the horizon that drives action, track useful warning time and compare with the validated schedule and a simple milestone-based predictor.
A forecast accurate one day before a delay is useful for immediate coordination but may be too late for a procurement recovery.
15:00 – 21:00
Cost on a common scope and Case 4
- In the book's notation, PV is the budget value of scheduled work, EV of recognised progress, and AC is cost incurred on a consistent basis; CPI = EV/AC, SPI = EV/PV. Earned value is not revenue, invoice value or cash, and excluding unpaid invoices from AC while the related progress is earned can overstate CPI.
Case 4 (assumed: BAC ₦10bn, PV ₦5bn, EV ₦4bn, AC ₦5bn) Recomputed CPI = 4 ÷ 5; SPI = 4 ÷ 5 0.8; 0.8 Variances EV − AC; EV − PV −₦1bn each; schedule variance is budget value, not cash or days EAC if past efficiency continues: 10 ÷ 0.8 ₦12.5bn; remaining cost ₦7.5bn EAC if remaining work runs at budget: 5 + 10 − 4 ₦11bn To-complete index at ₦11bn: (10 − 4) ÷ (11 − 5) 1.0 To-complete index to hold BAC: 6 ÷ (10 − 5) 1.2 Residual against assumed ₦13bn contract value ₦0.5bn or ₦2bn; neither is net profit - These are different assumptions, not competing truths; Chapter 19 notes a bottom-up estimate may instead find ₦6 billion of remaining cost, also ₦11 billion in total. The estimate to complete is the forecast cost of remaining work on a stated basis. An index of 1.2 asks the remaining work to beat budget efficiency by 20%.
- AT THE WORKFRONT · THE WORKED CASE
- Workfront. A fictional EPC contractor's site team knows the completed phase carried unusual mobilisation cost and the remaining quantities are well defined. That may support a forecast nearer ₦11 billion, but only through a bottom-up estimate.
IN THE COMMITMENT · THE PARALLEL- Commitment. The finance director sees a ₦1.5 billion gap between forecasts and a residual swinging from ₦0.5 billion to ₦2 billion. Nothing is accepted before bottom-up assessment and finance review; the decision is what the contractor does about the remaining obligation.
The formulas do not establish a claim, a cash need or accounting recognition; those need their own evidence.
21:00 – 25:00
The cost decision, and Pop Quiz 4
- Keep actual, committed, forecast and potential change amounts separate; adding a commitment and the actual it becomes double counts, and an unapproved variation is not revenue. Report earned progress, cost, commitments, remaining cost, revenue and cash separately, not as one green status.
- THE DOCTRINE
Each material forecast movement should have a cause, consequence and response. A forecast revised every month without action becomes a record of deterioration. Do not label an approved budget increase a saving.
- Today's assignment produces Section 18 — Delivery logic, stated assumptions, a common scope and the forecast of remaining work. Pop Quiz 4 opens now: five questions, closed book, on Sessions 16 to 18.
A healthy report makes the remaining obligation easier to understand, not easier to admire.
Key Terms
Total float
How far an activity can slip before it affects the finish in the represented network; not an unconditional right to consume time.
P80 date
The date by which 80% of simulated outcomes finish under the represented assumptions; not a guarantee of delivery.
Earned value
Budget value assigned to recognised progress under the agreed method; distinct from revenue and cash.
Estimate to complete
The forecast cost of remaining work on a stated scope and basis.
Self-Check — Before You Submit
1. In Case 2, B slips to day 39 and is expedited by three days, while C now finishes on day 37. What does the expedite recover against the day-52 forecast?
- A. Three days, finishing on day 49
- B. None, finishing on day 52
- C. Five days, finishing on day 47
- D. Two days, finishing on day 50
D. Two days, finishing on day 50 — B finishes on day 36 but D waits for C on day 37, so D runs to 45 and E to 50. Three days is the procurement-only claim the book warns against.
2. Case 4 gives BAC ₦10 billion, PV ₦5 billion, EV ₦4 billion and AC ₦5 billion. Which statement is correct?
- A. The −₦1 billion schedule variance is a cash loss
- B. CPI and SPI are both 0.8; holding BAC needs a to-complete index of 1.2
- C. EAC is ₦12.5 billion, so ₦11 billion is wrong
- D. SPI of 0.8 means 20% of the days are lost
B. CPI and SPI are both 0.8; holding BAC needs a to-complete index of 1.2 — 4/5 gives both indices; (10 − 4)/(10 − 5) = 1.2. The two EACs rest on different assumptions, and schedule variance is budget value, not cash or days.
3. A learned finish forecast is built on an update with missing logic and unsupported progress. What do you ask for first?
- A. Repair logic, constraints, lags, status and calendars before relying on any forecast
- B. Adopt it if its average error beats the schedule's
- C. A buffer on every activity
- D. Replace the schedule with the forecast
A. Repair logic, constraints, lags, status and calendars before relying on any forecast — A learned forecast on defective status hides the same defects behind a different number. Buffering every activity is tempting but double counts against a project contingency.
Today’s Assignment
Write Section 18 of your plan: the delivery logic, the stated assumptions and the forecast of remaining work behind one commitment in your company's decision chain.
- Draw the simplified network behind one milestone your company has promised (five to eight activities). Name the controlling path, the float elsewhere and what the network does not represent.
- Say honestly what is wrong with your company's schedule today.
- Test one recovery option in integrated form, as Case 2 does, with its safety, quality, resource and funding consequences.
- State the assumptions a risk simulation of this milestone would need, and what your P80 would and would not mean.
- Put cost and progress on a common scope for the same work (PV, EV, AC rules, cut-offs, accruals); compare an index-based estimate with a bottom-up estimate to complete and name your response. With no AI in use, work from the forecast in front of you.
YOUR 90-DAY PLAN
Section 18 — Delivery logic, stated assumptions, a common scope and the forecast of remaining work.
How this is marked — the rubric
| Presents a correct network with controlling path, float and stated limitations | 25 |
| Tests a recovery in integrated form with its safety, quality, resource and funding consequences | 25 |
| States explicit simulation assumptions and interprets the percentile correctly | 20 |
| Aligns cost and progress on a common scope and compares index-based and bottom-up forecasts with a named response | 30 |
| Total · pass mark 70 | 100 |
Submitting session 18
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: 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 — Readiness, Productivity, Schedule and Cost 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 4 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Schedule logic reviewCritical-path readingIntegrated recovery testingRisk-simulation interpretationEarned-value disciplineBottom-up estimate to complete
Part Four · Construction, Commissioning and AcceptanceSESSION 19
HSEQ and Work at Operating Facilities
Why This Matters
Nothing in this programme gives an AI service less room than the work that can hurt people. EPC work involves fabrication, lifting, excavation, electrical activities, testing and interfaces with hazardous operating facilities, and the arrangements for that work belong to competent assessment and the governing management system. Chapter 20 asks where intelligence genuinely helps: making evidence and accountability easier to see, reading leading indicators honestly and using cameras with their limits stated. The casebook then teaches a piece of statistics every leader should carry: what a perfect test score with zero failures does and does not prove. If your company works inside a client's operating facility, this session is about the boundary between you.
25-Minute Module
0:00 – 5:00
Intelligence supports the management system
- The applicable arrangements come from competent project assessment and the governing management system. The International Association of Oil and Gas Producers publishes recommended fabrication-site construction safety practices and its Process Safety Fundamentals as industry resources (IOGP, Fabrication Site Construction Safety Practices and Report 577; IOGP, Process Safety Fundamentals). They are references for consideration, not automatic permission for a particular activity.
- Map the contractor's responsibilities and the operator's. Identify who controls isolation, permits, access, simultaneous operations, emergency interfaces and return to service. The contract and site rules are assessed together.
- AI can retrieve approved information, compare a work pack with required evidence and identify unresolved actions.
- THE DOCTRINE
AI should not certify an isolation, approve a permit or infer a safe condition from a photograph.
If you cannot name who controls the isolation at the client interface, no service can help you yet.
5:00 – 10:00
Leading indicators with meaning
- A count of closed actions can conceal weak closure evidence. Track whether corrective actions were verified, whether critical conditions remain unresolved and whether recurring findings share a cause. Keep administrative closure distinct from physical correction.
- Near-miss and observation data need careful reading. Increased reporting can indicate stronger reporting behaviour rather than worsening conditions; a falling count can indicate underreporting. Examine exposure, reporting practices and actual findings before drawing conclusions.
- Predictive models trained on these data should state the population and collection limits. They can prioritise competent review.
- THE DOCTRINE
A low predicted score should not remove required controls. A model can prioritise competent review; it cannot grant an exemption.
When the near-miss count falls, ask first whether the work got safer or the reporting got quieter.
10:00 – 15:00
Cameras at the workfront, and environmental evidence
- A camera-based service needs an accepted purpose, coverage, privacy arrangements and a response process. It may detect a visible condition while missing context, authorisation or hidden hazards. Evaluate the lighting, weather, equipment and camera positions relevant to the intended use.
- False alerts consume supervisor capacity; missed alerts create false reassurance. Report both and examine consequential misses individually. Workers and supervisors keep the ability to act on what they see, whatever the model says.
- For environmental and waste evidence, connect the applicable requirements with actual work, measurements, waste movements and accepted records. An assistant can find inconsistent quantities and missing documents. It must not fabricate measurements or assume a disposal record proves an unverified activity.
- Use a consistent boundary for carbon or waste calculations and distinguish measured data from factors and estimates. A lower footprint caused by excluding a subcontract activity is not a demonstrated reduction; any external factor needs its own verified source and scope.
A camera that sees the hard hat cannot see the permit. Write down which one your decision depends on.
15:00 – 21:00
Case 8: what zero observed failures proves
- Suppose a service is tested and makes no errors. The question is how bad its true error rate could still be. The casebook assumes n independent comparable trials with a fixed failure probability p. The chance of seeing zero failures is (1 − p)^n. Setting that to 0.05 and solving gives p = 1 − 0.05^(1/n), a one-sided 95% upper bound under the assumptions.
Tests with zero failures Upper bound (recomputed) Reading 20 1 − 0.05^(1/20) ≈ 13.91% A failure rate near one in seven is not ruled out 100 1 − 0.05^(1/100) ≈ 2.95% Still not ruled out near three in a hundred 300 1 − 0.05^(1/300) ≈ 0.994% Roughly one in a hundred not ruled out - These are exact-model calculations rounded for presentation, not a rule requiring any test count for release. Actual AI failures can cluster in particular revisions, scans or question types, and a hard-case set is not a random sample from deployment. A thousand easy questions about clean single-version documents may say little about wrong-revision answers.
- Combine representative tests, which estimate performance for a population, with challenge tests, which probe known weaknesses, plus shadow operation and incident monitoring, keeping their purposes distinct. If a material error occurs in a critical category, investigate its cause and affected decisions even if the aggregate pass rate stays high.
- AT THE WORKFRONT · THE WORKED CASE
- Workfront. A fictional EPC contractor's HSEQ lead tests a work-pack evidence checker on 20 packs with no errors. Under the casebook's assumptions, that leaves an error rate of up to about 13.91% unexcluded, and none of the packs involved a withdrawn isolation.
IN THE COMMITMENT · THE PARALLEL- Commitment. The executive sponsor is asked to approve wider use. The release record states what the 20 tests support, what lies outside scope and which challenge cases are still owed. The service stays advisory; the permit authority is unchanged.
A perfect observed score is evidence about the tested cases. Ask how those cases were chosen.
21:00 – 25:00
The leadership decision
- THE DOCTRINE
If the work cannot proceed within accepted conditions, the useful output is the constraint and the required responsible action. Production pressure and a favourable financial forecast do not create permission.
- The intelligent HSEQ workflow makes evidence and accountability more accessible and preserves stopping and escalation arrangements. Judge it through better control and verified action, without inventing a percentage reduction in incidents that the evidence cannot establish.
- Write down one HSEQ indicator your company reports to its board and whether it measures verified correction or administrative closure.
- Today's assignment produces Section 19 — Support for the management system, meaningful leading indicators and their limits.
Judge an HSEQ service by better control and verified action, not by how many alerts it raised.
Key Terms
Leading indicator
A measure intended to show control before harm occurs; meaningful only when it tracks verified correction, not administrative closure.
Zero-failure upper bound
The highest failure probability not ruled out, at a stated confidence, after n independent comparable trials with no failures: 1 − 0.05^(1/n) at 95%.
Challenge test
A deliberately chosen set of hard cases probing known weaknesses; not a random sample from deployment.
Calculation boundary
The defined scope of activities included in a carbon or waste figure; excluding an activity is not a demonstrated reduction.
Self-Check — Before You Submit
1. A vendor reports 100 tests with zero failures. Under the casebook's assumptions, what can you say?
- A. The service has a 0% failure rate
- B. A true failure rate of up to about 2.95% is not ruled out at one-sided 95%, if the trials are independent and comparable
- C. The service is proven safe for release
- D. The failure rate is exactly 2.95%
B. A true failure rate of up to about 2.95% is not ruled out at one-sided 95%, if the trials are independent and comparable — 1 − 0.05^(1/100) ≈ 2.95% is an upper bound under stated assumptions, not the rate itself and not a release rule. Reading zero observed as zero true is the trap.
2. Your monthly near-miss reports rose sharply after a reporting campaign. A director calls it a deterioration. What is the sound response?
- A. Agree and tighten site discipline at once
- B. Ignore it, since near-miss data are unreliable
- C. Examine exposure, reporting practice and actual findings before concluding; more reporting can mean stronger reporting behaviour
- D. Train a model to predict the next near miss
C. Examine exposure, reporting practice and actual findings before concluding; more reporting can mean stronger reporting behaviour — The book says increased reporting can indicate stronger reporting rather than worse conditions. Dismissing the data is the opposite error; the findings still need review.
3. A site camera service has a strong detection score. A supervisor wants to remove a manual check in areas the model scores as low risk. What do you decide?
- A. Keep the required control; a low predicted score does not remove required controls, and the model may miss context, authorisation or hidden hazards
- B. Approve, since the score is strong
- C. Approve for night shifts only
- D. Approve if the vendor reports zero missed alerts
A. Keep the required control; a low predicted score does not remove required controls, and the model may miss context, authorisation or hidden hazards — A model can prioritise review but not remove required controls, and visible detection misses hidden hazards. A vendor's zero misses is itself only evidence about its tested cases.
Today’s Assignment
Write Section 19 of your plan: how intelligence supports your company's HSEQ management system for one decision in your chain, which leading indicators carry meaning and what the limits are.
- Name the HSEQ-relevant point in your decision chain and map who controls isolation, permits, access, simultaneous operations, emergency interfaces and return to service, contractor and operator each. Clients: write it from the operator's side.
- State honestly which indicators your company reports today and whether they measure verified correction or administrative closure.
- Define two leading indicators with meaning, with the evidence of verification and how reporting behaviour will be read.
- If a camera or evidence-checking service is proposed, state its purpose, coverage, privacy and response arrangements, how false and missed alerts are reported, and what it cannot see.
- State the test evidence you would require before wider use, using the zero-failure bound to say what a perfect score would and would not support, and what stays outside scope.
YOUR 90-DAY PLAN
Section 19 — Support for the management system, meaningful leading indicators and their limits.
How this is marked — the rubric
| Maps contractor and operator responsibilities and keeps permit and isolation authority with people | 25 |
| Defines leading indicators that measure verified correction and interprets reporting behaviour honestly | 25 |
| States the limits of camera and environmental evidence, including false and missed alerts and calculation boundary | 25 |
| Specifies test evidence correctly, using the zero-failure bound and separating representative from challenge testing | 25 |
| Total · pass mark 70 | 100 |
Submitting session 19
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: 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
Management-system alignmentMeaningful leading indicatorsWorkfront vision limitsEnvironmental evidence boundariesZero-failure reasoning
Part Four · Construction, Commissioning and AcceptanceSESSION 20
Commissioning, System Completion and Handover
Why This Matters
This is where your company is finally judged: not on how much it installed, but on whether the client receives a usable system with evidence that holds together and a commercial position that can be invoiced. Chapter 21 asks you to plan backwards from that usable system, to agree what each completion state means, and to keep evidence and configuration together so a test still covers what is actually installed. The casebook shows why a package three-quarters complete can be zero per cent ready, and the walkthrough closes on a source mismatch the assistant must not resolve by itself. It is also the midpoint of the programme, and Practice Exam 4 opens at the end.
25-Minute Module
0:00 – 5:00
Plan backwards from a usable system
- Commissioning is not a final administrative stage after construction. Define system and subsystem boundaries, completion sequence, tests, interfaces and acceptance evidence early enough to shape engineering, procurement and packaging.
- A contractor can install large quantities while a system cannot progress because of a small component, an unresolved logic change or a missing accepted test record.
- AI can compare the completion plan with evidence and produce a gap list naming object, condition, source, state and owner, and distinguishing missing evidence from evidence of a failed condition.
- Define the project's actual meanings for mechanical completion, precommissioning, commissioning readiness, energisation, introduction of process fluids, performance testing and final acceptance where they apply; terms and boundaries vary by agreement.
- THE DOCTRINE
Do not assume that one signed certificate authorises every later activity. Each transition requires its own accepted conditions and responsible authority.
Ask your commissioning lead to name the smallest item that could stop your system. That is where your attention belongs this month.
5:00 – 11:00
Case 1 and controlled state transitions
- The casebook's hypothetical control package has four enabling conditions: approved configuration; panel received and accepted; field connections accepted; required functional evidence accepted. Three are complete and functional evidence is pending, so an average score reads 3 ÷ 4 = 75%.
- Each accepted state g1 to g4 equals one only when the responsible authority has accepted current evidence. Readiness is R = g1 × g2 × g3 × g4. With states 1, 1, 1, 0, R = 0. The multiplication represents mandatory conjunction, the same rule as Session 16, at system level; it is not a probability model.
- If the panel receipt was accepted before a configuration change, it may remain valid physically while compatibility needs reassessment: examine the affected conditions rather than deleting all evidence or assuming it all applies. Completed conditions may be reported as progress beside readiness, correctly named.
- A completion service represents the project's accepted states and transitions; a generic template cannot settle them. It separates an absent record, a failed test, a pending review and an accepted result, each resolved by a different owner. Deterministic rules run through a reviewed rules service; the language model finds and explains gaps.
- THE DOCTRINE
Do not populate an acceptance field directly from generated text. No percentage of found files can substitute for a missing mandatory condition.
Progress helps plan the work. Readiness decides whether the next stage may proceed. Never let one wear the other's name.
11:00 – 16:00
Evidence and configuration belong together
- Handover establishes the installed configuration, accepted departures, test results, outstanding conditions and usable operating information. Link serial numbers, software and control versions, drawings, procedures and test records. A test on an earlier configuration may need reassessment after a change, and the technical authority decides which evidence still applies.
- An AI check can compare tags, serial numbers, revisions and software identifiers and report the exact mismatch with sources. If a scanned value is ambiguous, it requests review rather than choosing the likelier serial number, and it does not merge as-built drawings, vendor manuals and inspection records into one apparent truth.
- The book's assumed walkthrough closes here. Before stage T, the assistant assembling evidence for S-01 finds a test record linked to revision A while the installed configuration register shows revision B. It cites both and requests technical assessment. It does not declare the system failed, nor assume the earlier test still covers the change.
- The reviewer determines what evidence remains applicable and what further work is needed; the assistant's next brief rests on that accepted disposition, not its earlier suggestion.
When a test and the installed configuration disagree, the assistant's job is to show both sources, request technical assessment and stop there.
16:00 – 21:00
A usable index, guarantees and the commercial evidence
- Agree the receiving organisation's information structure early: system, equipment, accepted configuration, maintenance information, spares, test evidence, outstanding conditions and the transfer record, with access and reuse rights. A document called final may contain a preliminary drawing. Keep unresolved items in a maintained register, never removed to make the package sound complete.
- Where the contract includes performance tests, establish the agreed conditions, measurement, responsibilities and treatment of departures. A model predicting performance does not replace the accepted test, and a successful isolated test may not satisfy all contract conditions.
- Map the contract's invoice events to their evidence; technical completion and payment eligibility can fall at different times. Do not extend a subsystem certificate to another because the names are similar, and keep any acceptance condition in the commercial brief.
- AT THE WORKFRONT · THE WORKED CASE
- Workfront. In the walkthrough, the receiving team tests the S-01 handover index by locating the accepted configuration, test basis, operating information and remaining conditions. A searchable archive that retrieves the wrong revision quickly has failed the practical test.
IN THE COMMITMENT · THE PARALLEL- Commitment. Commercial leadership separately checks which accepted records support the invoice milestone. The service prepares the pack; the authorised owner approves the submission. A later client request for a missing document is routed to the right owner; it does not automatically mean the installed work failed.
Work can be technically complete while the contractual position is unresolved. Name which one you have before you name the owner.
21:00 – 25:00
Measure the finish, and Practice Exam 4
- Report systems ready for the next accepted stage, conditions preventing progression, overdue evidence and decisions required. Track time from completed work to accepted evidence and from acceptance to invoice eligibility. At handover, ask receiving users to find the evidence for a selected object and explain its remaining conditions.
- Before using a completion assistant's conclusion, test it on correct transitions, missing records, superseded results, duplicated certificates, wrong equipment identity, unresolved punch items and changed conditions, at the same cut-off. Examine every false claim of accepted readiness in a critical category, and keep a fallback to authoritative records.
- THE DOCTRINE
Measure the finish rather than the file count. A mass of files without searchable identity or relationships can satisfy a crude metric while creating a poor operating handover.
- Today's assignment produces Section 20 — The usable system, accepted completion states and evidence kept with configuration. Practice Exam 4 opens now: fifteen questions on Sessions 1 to 20, closing Week 4.
Handover is too important to depend on an index nobody can reconstruct.
Key Terms
Completion state
A project-defined stage such as mechanical completion or energisation, with its own accepted conditions and responsible authority.
Readiness function
The casebook's R = g1 × g2 × g3 × g4: one only when every accepted state is one. A conjunction, not a probability.
Installed configuration
The accepted physical and documentary state actually in place, against which tests and handover evidence must be reconciled.
Invoice eligibility
The point at which the contract's invoice event is supported by its required accepted evidence; it may differ from technical completion.
Self-Check — Before You Submit
1. In Case 1, three of four enabling conditions are accepted and functional evidence is pending. What is the readiness, and what may management report?
- A. 75%, so the next stage may proceed with a punch item
- B. 1, because the evidence is expected soon
- C. 0; completed conditions may be reported as progress, with the blocker shown
- D. Unknown until a model predicts the pending result
C. 0; completed conditions may be reported as progress, with the blocker shown — R = 1 × 1 × 1 × 0 = 0. The 75% average is progress, not permission, and the missing state must not be filled with a model prediction.
2. The completion assistant finds a test record on revision A while the installed configuration register shows revision B. What should it do?
- A. Mark the system failed
- B. Cite both records and request technical assessment, without declaring failure or assuming the earlier test still applies
- C. Assume the test still applies because the change was minor
- D. Choose the revision with more supporting documents
B. Cite both records and request technical assessment, without declaring failure or assuming the earlier test still applies — The technical reviewer decides which evidence remains applicable. Declaring failure and assuming coverage are both conclusions the assistant has no authority to reach.
3. Your commercial lead wants to invoice a subsystem milestone using a certificate issued for a similarly named subsystem. What do you decide?
- A. Invoice; the names show they are related
- B. Invoice and correct later if queried
- C. Hold all the project's invoicing until handover
- D. Map the invoice event to its own accepted evidence and keep any condition in the brief
D. Map the invoice event to its own accepted evidence and keep any condition in the brief — The book warns that extending a certificate by similar names can create both a handover error and an unsupported invoice. Holding all invoicing overcorrects; the issue is this milestone's evidence.
Today’s Assignment
Write Section 20 of your plan: the usable system at the end of your company's decision chain, its accepted completion states and how evidence is kept with configuration through to invoice.
- Name the system or deliverable your decision chain ends in and plan backwards from it: boundaries, completion sequence, tests, interfaces and acceptance evidence. Vendors and consultancies: use the accepted deliverable your client receives.
- List the completion states your company uses and say honestly where their definitions are assumed rather than agreed, and where one certificate is treated as authorising later activities.
- For one transition, list the mandatory conditions as accepted states and show the readiness function, separating absent, failed, pending and accepted.
- State how test evidence is linked to installed configuration and who decides applicability after a change.
- Map one invoice event to its required evidence, separate performance guarantees from commercial acceptance, and replace one file-count metric with a measure of the finish.
YOUR 90-DAY PLAN
Section 20 — The usable system, accepted completion states and evidence kept with configuration.
How this is marked — the rubric
| Plans backwards from a named usable system with boundaries, sequence, tests and acceptance evidence | 25 |
| Defines completion states with conditions and authorities, and admits where definitions are assumed | 25 |
| Keeps evidence with configuration and leaves applicability decisions with the technical authority | 25 |
| Maps an invoice event to accepted evidence, separates guarantees from acceptance and measures the finish | 25 |
| Total · pass mark 70 | 100 |
Submitting session 20
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: 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.
Sit Practice Exam 4 →Closed book, no notes, and Coach Ritchie switches himself off while you sit it.Capabilities Covered
Backward completion planningCompletion-state definitionConfiguration-linked evidenceUsable handover indexAcceptance-to-invoice mapping