One illustrative onshore gas development, 500 MMscfd export compression, eight train layouts, answered at the depth each FEL stage can honestly support. Start in Base; switch to Advanced for every input and check.
Illustrative inputs only. Failure, repair and event figures are round screening values chosen to show the method, not OREDA, vendor or risk-study data.
Base view keeps one question: which train layout, and does the answer hold? Advanced view adds every input, the FEL 1 screen, five more robustness checks (sensitivity, major events, phasing, spares, design margin), the FEED simulation and the formulas.
How to read the groups: Decision is what an investment board compares between options. Production is what operations plan around in the selected year. Investment and Exposure explain where the decision number comes from. Everything in the tabs below either feeds these numbers or tests whether the decision holds.
Compression CAPEX per train follows the six-tenths rule: train cost = CAPEX(100%) x size^0.6. A train built later is counted at its present value. Deferment is valued as revenue lost in the year it occurs; volume recovered after field life can be credited with the last input.
Design margin over plateau. No API, ISO or NORSOK standard fixes one plant-level margin over plateau. NORSOK P-100 (cl. 4.2.4) makes the design margin and turndown the project owner's responsibility, and ISO 20815 with ISO 15663 expect the choice to be justified by production assurance and lifecycle cost. Equipment standards give the anchors: NORSOK P-002 sets at least 10% over design rates for scrubbers (P-100 cl. 5.3.2: 20% for process-control fluctuations); API 675 rates metering pumps at no less than 110% of maximum specified flow; API 617 sizes compressor drivers at least 10% above maximum absorbed power; API 610 leaves the rated-to-normal flow margin to the purchaser. Process licensors commonly add 10% on flow, and stacked margins are a known cause of oversized compressors (Kurz, Thorp, Zentmyer; Solar Turbines and Saudi Aramco). Default here: 10%.
Train count and size. NORSOK P-100 classes each system by the consequence of its failure for deliveries: A, shutdown or reduced production for more than 2 hours; B, less than 2 hours; C, no effect. Gas compression is class A, "vital for production". For some class A services, such as export pumps, the standard requires a spare unit or spare capacity; for compression it leaves the sparing to the project and does not choose between 2 x 100%, 3 x 50% or 2 x 60%: that choice is the regularity and lifecycle-cost optimisation this page runs (ISO 20815, ISO 15663). The configuration table shows which layouts give full N+1 at the chosen margin and the train size that gives full N+1.
Up to eight train layouts. Train size is a share of the plateau rate before design margin; installed size = size x (1 + design margin). Set the number of trains to 0 to leave a slot out. Layouts whose installed capacity exceeds plateau show what extra or larger trains buy and what they cost. N+1 test: with one train out, the remaining trains still carry plateau rate.
Which options go forward, and which structural weaknesses they carry.
Benchmark ranges only. No vendor, no equipment list.
Does the screening survive the full range of benchmark and cost uncertainty?
Options share one benchmark set and one estimating basis, so their errors move together. An option is dropped only if another is cheaper in all four corners (best and worst availability, low and high CAPEX accuracy). The test prices cost and deferment only: check what a dropped option offers that it does not price, such as a lower turndown floor.
Lifecycle cost of each carried option at base inputs, and what drives the difference.
Generic failure data per block, plus the interfaces that operations own.
Does the winner survive a different gas price or a CAPEX overrun?
Each panel asks one question: does the ranking above still hold if this changes?
Each input set to x0.5 and x1.5 of its base value, one at a time. Inputs at the top deserve the most effort in FEL 2 data gathering.
Random failures last hours. Fire, explosion or external damage lasts months and takes out an area. The event removes one train area (the whole plant if there is one train). Repair cost is excluded. An event reaching several trains or a shared utility is worse than shown.
Change the deferral in Facility at a glance. Phasing pays when the deferred train adds capacity the field may not need yet; it costs when the deferred train is the N+1 reserve.
Lifecycle cost of each carried option at three design margins. Margin makes every multi-train layout partly self-sparing: with one train out, the others carry more than their nominal share. For a single train it buys only catch-up capacity, which this model does not credit.
How the focus option's numbers are built.
R(t) = e^(-t/MTBF): chance of running t hours without failure. A = MTBF / (MTBF + MTTR): share of time the train can run. M(t) = 1 - e^(-t/MTTR): chance a repair is finished within t hours.
The range of plateau years the focus option will actually deliver.
Maintenance strategy: crews, repair scatter and spares logistics.
Distribution of simulated plateau years.
Plateau production efficiency of the focus option as each stage would report it. FEL 1 and FEL 2 ranges describe what is not yet known about the inputs. The FEED band describes how much one operating year differs from the next.
Method follows ISO 20815 (production assurance and reliability management), ISO 15663 (life cycle costing) and the NORSOK P-100 / P-002 system sparing classes; block relations as in IEC 61078 and IEC 61703. All calculations run in this page.