- Home
- Case studies
- Coastal refiner
A coastal refiner: one capital book across four decisions.
A coastal refiner, invented for this page, owns an integrated refinery and petrochemical complex. This year it is choosing new units to build, backing climate-technology ventures, recovering a programme in construction and running its annual capex round. Four teams work from four spreadsheets, but they draw on one $600M envelope and share one margin cycle. This is the same invented book the home page and the capital committee page walk through, told here in full, in US dollars.
On screen
One assumption replaced, then one envelope
The workspace for this example, reconstructed. A flat supply assumption gives way to the tracked capacity pipeline, the expansion plan re-solves, and the committee sees all four decisions on one plan.
Every figure is invented for this example. In a real workspace the inputs come from the files and systems your teams already use, and the approved plan goes back to them as budgets.
The situation
Four decisions drawing on one pool of money
Every figure on this page is invented. The refiner's capital committee has a $600M envelope for the three years FY27 to FY29, and an appetite for a bad year: the book's value after hurdles, averaged over the worst 1 in 20 scenarios, must stay above $700M. Four teams bring claims on it:
- Expansion. Seven candidate units: a polypropylene unit, a paraxylene unit, a hydrocracker upgrade, a butadiene extraction unit, a specialty-solvents train, a polyols plant and a propylene splitter.
- Ventures. A corporate venture programme in climate technology with its own $120M over four years, and two strategic stakes too large for it: 20% of an offshore wind project and a 10 MW hydrogen pilot.
- Construction. A five-unit programme already under way, $950M at award, led by the crude unit revamp.
- Sustaining capex. 112 small lines against a $150M sustaining budget, and two lines too large for it: a 132 kV grid substation and a 40 MW solar captive plant.
All four are exposed to the same things: crude and gas prices, an aromatics and polymer cycle under pressure from new regional capacity, copper, one transformer supplier and the same contractor crews. Funding each list on its own ignores all of that.
What Capibud does
Each decision on shared scenarios, then one committee
- Evaluate: seven units, solved together. The units are simulated on the same 4,000 scenarios as the rest of the book, across 16 shared factors, with hard limits that a ranked list ignores: about 210 kt a year of spare reformate for the aromatics units, one plot that can take the polyols plant or the specialty-solvents train but not both, and a paraxylene unit that needs the hydrocracker upgrade first. Four synergies are priced in, such as a shared jetty and tank farm for the paraxylene unit and the hydrocracker upgrade, and the propylene splitter feeding the polypropylene unit. The planning case builds the polypropylene unit now, the hydrocracker upgrade next and the paraxylene unit after it: $583M of P50 value, with a bad year of −$196M.
- Evaluate: one assumption replaced. A flat regional paraxylene supply assumption of 1,250 kt a year is replaced with the tracked regional capacity pipeline, 1,780 kt a year by 2032. On the same plan, P50 value falls from $583M to $489M and the bad year from −$196M to −$262M. The paraxylene unit's case now reverses under a 0.6σ fall in aromatics margins, a move that comes about one year in four, where the planning case needed 1.8σ. Re-solved, the plan drops the paraxylene unit and the hydrocracker upgrade, which only cleared its hurdle as the paraxylene unit's prerequisite, and brings the butadiene extraction unit forward to 2029: $521M of P50 value, with a bad year of −$171M.
- Evaluate: the unit that goes to committee. The polypropylene unit is this round's only FID. Two trains cost $1.12bn at P50 and $1.31bn at P80; the refiner's equity over the three years, its claim on the envelope, is $264M at P50 and $301M at P80, and project debt carries the rest. The plan says March 2029; the P50 finish is June 2029 and the P80 November 2029. The full build is worth $412M at P50. Staging train 2 gives up $57M of that and is $96M better in the bad year, and deferring a year is worth $318M.
- Invest: the venture book on the same factors. Inside its own $120M, the programme backs five companies with $42M this round and holds $38M of reserves against a P80 follow-on need of $35M, with no single name above 12%. A pilot-plant ticket held over from last year is released: after the six-point venture premium, its $18M does more elsewhere. Strategic value is priced at the scenario carbon price, so a carbon shock moves the ventures and the decarbonisation capex together. The two stakes above the programme's limits go to the committee: the wind stake at $72M in two tranches, and the hydrogen pilot at $66M.
- Execute: the monthly review. Five units, from the crude unit revamp to the product jetty, are forecast from actuals. Mean forecast cost comes out $34M above budget authority. At P80 (the cost stays below this in four scenarios out of five) the overrun is $68M, and at P90 it is $91M. Contingency of $26M covers 38% of the P80 requirement. The crude unit revamp sets the programme finish, and the team brings one claim to the committee: recovery option B, which funds a second mechanical and piping shift, a crew moved from the jetty and tie-ins resequenced into the turnaround, together with the budget to carry the revamp to completion. It asks $38M at P50 and $45M at P80, and is worth $41M at P50 against accepting the slip.
- Capex: the sanction round. 81 of the 112 lines are fitted to the $150M sustaining budget, the crews and the third-quarter turnaround, with all 14 statutory deadlines met. 31 lines wait, at an expected loss of $9.4M, and the funded lines go out as 25 proposals: 13 to plant managers, 8 to the CFO and 4 to the executive committee and board. The substation ($96M) and the solar plant ($52M) are too large for the sustaining budget and go to the committee. A separate example, the process-plant capex round, covers a round like this in detail.
- Committee: one envelope. The six claims ask for $588M at P50 and $664M at P80, against $600M. They are pooled on 4,000 shared scenarios (run 7f3a…c21e, seed 4417, engine 3.2). Each carries a premium for its risk class over a 9% real base hurdle (an 11.25% cost of capital in money terms): four points for a new unit, six for a venture stake, one for a capex line and none for recovery on a committed programme. The book is solved for the most value after hurdles with its bad year above $700M.
Throughout, Capibud sits on top of what the four teams already use. It reads their models, cost reports and request sheets, decides where the money goes, and sends the approved plan back to them as budgets, re-phasing and tripwires.
Results
A better plan on less money, from two deferrals
| Method | Expected value | After hurdles | P10 | Bad year1-in-20 | Capital |
|---|---|---|---|---|---|
| Pooled plan (4 claims) | 1,412 | 1,180 | 896 | 815 | 470 |
| Today's separate plans (6 claims) | 1,396 | 1,139 | 842 | 761 | 588 |
| Profitability-index ranking (5 claims) | 1,371 | 1,122 | 788 | 664 Below floor | 560 |
Illustrative example — figures invented for illustration; not a client or a Capibud demo. The pooled plan holds the remaining $130M of the $600M envelope in reserve, released by the committee. At P80 it commits $520M.
Where the gain comes from
Today's separate plans fund all six claims: each team funds its own list. The pooled plan funds four and defers two, and the difference is worth $41M after hurdles and $54M in the bad year, on $118M less capital:
- Defer the hydrogen pilot a year: +$29M after hurdles, +$31M in the bad year, $66M freed. Its $9M of P50 value does not cover the six-point venture premium, and waiting a year keeps the option to build it if the offtake firms up.
- Defer the solar captive plant with it: +$12M after hurdles, +$23M in the bad year, $52M freed. Its case rests on powering the electrolyser, and it shares copper and the transformer supplier with the substation and the wind stake. Waiting costs it a little value at P50, which the option more than repays.
A profitability-index ranking funds five claims and does worse on both counts: its bad year of $664M falls below the floor, because it takes three copper-exposed claims at once.
The polypropylene unit, full build
Staging train 2 would give up $57M at P50 to be $96M better in the bad year. With the book's bad year $115M above the floor, the committee keeps the full build and arms staging as a tripwire: if the polymer spread falls below $310/t, about 1.4σ ($62/t) under its $372/t mean, the decision re-opens with a staged re-run drafted. At $264M the approval sits inside the CFO's $300M limit, so decision pack DP-0412 is approved by the CFO and the board is noted.
The programme in construction
| Unit | Budget at award | CPI · SPI | P80 finishvs target | Verdict |
|---|---|---|---|---|
| Crude unit revamp | 310 | 0.92 · 0.90 | Nov 2028 · Apr 2028 | Escalate |
| Low-density polyethylene line | 245 | 0.96 · 0.94 | Feb 2028 · Nov 2027 | Recover |
| Cogeneration and utilities | 205 | 0.97 · 0.98 | Dec 2027 · Oct 2027 | Watch |
| Sulphur recovery block | 105 | 1.01 · 0.99 | Sep 2027 · Sep 2027 | Proceed |
| Product jetty and pipelines | 85 | 0.99 · 1.02 | Aug 2027 · Aug 2027 | Proceed |
Illustrative example — figures invented for illustration; not a client or a Capibud demo. The five budgets sum to the $950M programme.
The crude unit revamp sets the programme finish, about seven months beyond target at P80, and each month there costs about $5.8M of value. The contractors' latest milestone dates run 30 to 65 days earlier than Capibud's P80, four change orders are pending, and the revamp contractor's CPI is 0.86. Its EPC crew, crew A, also builds the polypropylene unit, so a slip on one moves the other in the same scenario. Recovery option B is funded; option C, a second contractor, is held back and re-opens if crew A slips by more than four months. Crew levelling allows for the turnaround's draw on mechanical crews and a monsoon derate, each labelled as an estimate.
What would break the plan
Of five named stresses, only a combined downturn (copper, polymer spread and crews together) takes the bad year below the floor, to $662M. The reverse stress test finds the smallest shock that does: copper up 46% together with a five-month slip for EPC crew A, about once in 35 years on the calibrated history of both. The nearest single factor is a 41% fall in the polymer spread. The plan's value after hurdles falls to zero only under a joint factor move of about 3.1 standard deviations, led by petrochemical and refining margins. Its advantage over today's plans disappears if the carbon price rises by more than about 1.7 standard deviations, because the hydrogen pilot and the solar plant would then be worth building now; the brain watches that threshold.
Limits
Read the gain for what it is
- The figures are invented. Every number on this page was made up to show how the method works. It is not a client, not a Capibud demo workspace and not drawn from any company's data.
- The size of the gain is not a forecast. $41M on $1,139M is about 3.6%. Pooling helps most where decisions share money and risk; in your portfolio the gain could be larger, smaller or nil.
- Hurdles drive the answer. Deferring the hydrogen pilot follows from the six-point venture premium. A committee with a different policy would get a different plan, and should.
- No outcome evidence. Nothing here shows that a pooled plan does better in reality. That needs real decisions and time.
What this means for you
Most companies already have the four lists
Shared constraints
If your projects compete for feedstock, plot space, a grid connection or crews, a ranked list can recommend a plan you cannot build.
Shared exposure
If one cycle drives several decisions, it should hit them in the same scenario. Then the bad year means something.
One table
Recovery money, venture stakes and capex lines compete for the same envelope. The committee should see them together.
Related
Use cases and industries in this example
Every decision in this example has its own guide. The refining and petrochemicals packs supply the kind of baseline it would start from.
A process plant's capex round
64 projects, statutory deadlines and a crew peak at 109% of capacity.
Read the exampleEvaluate: sanction and stage-gates
Options, sizes and timing on shared scenarios.
Read the guideExecute: projects in construction
P80 cost and finish from actuals, recovery priced in money.
Read the guidePool your four decisions on one envelope.
A Decision Sprint runs one portfolio on your own files in six weeks, with success criteria agreed in advance.