Read the complete supplied case · BPB2-19
# BPB2-19 — Solar farm construction — use previous experience intelligently
Suite BPB2 v0.3-pilot; Foray 247, third essai.
### Scenario
Merefield Energy is preparing a forty-eight-megawatt solar farm on coastal clay. Project manager Eva Chen has an apparently persuasive lessons pack, but its headline savings mix crew numbers, weather, off-site effort and different design conditions. She must propose concrete plan changes on 22 March 2027 before investigation and purchasing windows close. Preventive money and engineering attention are limited. Eva needs to decide what previous experience warrants committing now, what should be tested, and which attractive lessons should not enter this baseline.
### Exact task prompt
Produce one usable in-browser answer page for the named practitioner, using the supplied fictional evidence. Keep observed facts, forecasts, assumptions and proposed decisions distinguishable, and cite source IDs beside material claims. No external research or real-world certification is required.
Prepare Eva's evidence-to-action decision memo. Select a coherent package of transferable lessons and translate them into specific funded commitments, baseline amendments and controls for this site. Audit the strongest claimed improvement rather than accepting its headline. State the mechanism and boundary conditions behind each transfer, identify a credible competing explanation, and define what observation would cause the action or plan to change. Show an executable approval/timing route and one defensible alternative package. Do not fabricate probabilities, expected savings or causal certainty; equally, do not defer every action until perfect evidence exists.
### Supplied records
All projects and records are synthetic. Current decision is D1=Monday 22 March 2027; D numbers count weekdays, no holidays. Work ofn days starting Ds occupies Ds through D(s+n−1); its report is usable at that day's end. An order/investigation may start the workday after its approval. Currency GBP; historical nominal costs may support mechanism/context analysis but must not be treated as current quotes.
**19-A — Current baseline/authority, director Omar Shah, D1.** Site 48MW, four ground zones of twelve pile blocks each. Coastal clay, intermittent cobbles in two zones; existing test coverage is two shallow desk-study points, insufficient to map the cobbles. One low-lying access road, no alternative. Foundation order is released at the **start of D18**; any change must use evidence available by end D17. Design freeze D25; main works/haulage D31. Grid-interface details are expected D20; the operator will not approve them earlier. Remaining risk allowance £150k. Up to £95k may be committed to prevention, but Omar's cumulative delegated limit is £75k; board approval above that is available at its D5 meeting. Separable investigation can be approved before the rest of a package; artificial splitting of one order is forbidden. Gia Cole has eight engineering days available, flexibly schedulable within D2–D17; other engineering demand already fills the rest of her capacity. The quoted consecutive analysis/report periods must remain consecutive. Leon Ward owns logistics; Fay Moss owns the integrated plan.
**19-B — Programme newsletter, 2026, sponsored by the former delivery director.** “Prefabrication cut installation from 12 weeks to 8: a 33% efficiency gain, with an 8% steel saving. Repeat this standard. Trial piles and access roads are optional risk costs; the real lesson is to order kits early.” The article compares L1 and L2 below; its cost claim refers to L5. No further calculation is attached.
**19-C — Signed close-out summaries and diaries, retrieved 17 March.** The recorded weeks cover on-site installation from crew arrival to completion, including stoppages; five days/week. Crew counts stayed constant. These are completed scope, not planned MW.
| Source | Scope/method/resources | Observed outcome and conditions |
|---|---|---|
| L1 Wren 2024 | 48MW conventional;2 crews;12 on-site weeks | Two full weeks stopped by wet access; inland clay; no pretrials; refusal problems concentrated in one cobble band; foundation changes cost £180k |
| L2 Kestrel 2025 | 48MW prefabricated;3 crews;8 on-site weeks | Chalk, dry; no weather stoppage; design stable before kit order; ten off-site fabrication weeks precede the measured interval and are excluded; storage/transport cost £70k excluded from the claimed steel saving |
| L3 Lark 2026 | 40MW coastal clay; access reinforcement £42k | Zero access closures; unusually dry season; planner had allowed five closure days, but no observed untreated comparator exists |
| L4 Swift 2026 | 44MW; trial piles £26k in six selected blocks | Detected redesign before orders in those blocks; refusal later occurred in two untested zones. Director claimed £120k “avoided”; no priced counterfactual survives |
| L5 Osprey 2025 | 32MW; five-project pooled steel order |8% quoted steel discount, but £50k storage damage; other projects provided most purchasing volume; contract prices do not isolate a prefab effect |
| L6 Rook 2026 | 48MW prefabricated;2 crews;10 on-site weeks | Design changed after kit order;£95k modification plus three weeks off-site rework before crew arrival, excluded from on-site measure; same supplier as L2 |
**19-D — New-site evidence, Gia/Leon, D1.** The two suspected cobble zones are non-adjacent, so six centrally selected test positions would not represent the site. The road's drainage survey confirms a low point that would prevent heavy deliveries when flooded; no probability or closure duration estimate is validated. Operator interface details may change string grouping and associated kit labels, but foundation geometry can be fixed independently after ground results. The project has neither a five-project buying framework nor a secure full-kit holding area. These are current facts, not an assurance that any intervention eliminates delay.
**19-E — Fixed current offers, D1; all costs fully inclusive.**
| Action | Cost | Duration, resources and limits |
|---|---:|---|
| G8: eight targeted trial locations |£18k | Six workdays fieldwork then four analysis/report days; Gia required for all four analysis days; field crew independent. Land access allows fieldwork only D1–D15 |
| G16: sixteen distributed trial locations |£30k | Ten fieldwork days then four analysis/report days; same Gia requirement and access rule; mutually exclusive with G8 |
| R: road drainage/reinforcement |£38k | Five workdays, can be done D26–D30 if ordered by D20; no Gia days; must finish before D31 heavy traffic |
| I: interface/kit boundary workshop and controlled design split |£12k | Four Gia-days before D17; does not obtain early operator approval. Produces a controlled separation of independent foundations from unconfirmed kit details |
| K: full-kit procurement/storage/expediting |£55k incremental | Needs approved interface freeze by D10 to meet baseline; no supplier substitute can relax this. Also needs six Gia-days by D17 |
| P: small reversible kit-label trial |£8k | Two Gia-days by D17; can test change-handling and traceability without committing full kits; no guaranteed programme saving |
No action earns an automatic cash saving or releases risk allowance beyond the amount actually unspent. The groundwork quotes cover testing/reporting, not subsequent redesign or remediation. Programme effects of discoveries must be reassessed. Purchases have not been made.
**19-F — Controls and decision rights, Omar, D1.** Eva may recommend the package and propose baseline/control changes. Gia must choose and document the sampling design; do not invent technical design or safety certification. Omar accepts findings into the project baseline; the operator retains interface approval. If ground results cannot support the D18 order, an explicit hold/partial-release decision is needed; assumptions are not accepted ground evidence. A staged package is permitted, with exposure and the later board decision visible. Keep a named owner for each trigger and evidence item.
### Browser deliverable and completion conditions
Produce a navigable decision memo linking each selected/rejected lesson to source evidence, competing explanations, the current-site mechanism and a specific baseline/control change. Show budget, engineering-capacity and approval/deadline implications for recommended and alternative packages. Selecting G8/G16 and optional measures must update those checks; source disclosures must retain adverse evidence, not just the favourable quotation. The practitioner should be able to see exactly what is authorised/proposed now and what happens if investigation results are unfavourable or late.
The page must work in the supplied browser/runtime; a download button or file preview alone is insufficient. Include a clear “Return to gallery” link using the shared shell route, or ../index.html if no route is supplied. Required controls must change the dependent results, not just their appearance. Original evidence and the initial dated position must remain recoverable; hypothetical changes do not create approvals or alter source records. Optional exports are secondary.