ashfordeOU Life-Cycle Cost Estimation (vehicle-design/cost-estimation/life-cycle-cost)
Use when the task is a full life-cycle cost (LCC) estimate for an aircraft program: the RDT&E, production, operations and support, and disposal phases, present value discounting of future year dollars, inflation escalation, and an uncertainty bound on the total.
Domain quick reference
- LCC phases: RDT&E (non-recurring: design, prototypes, test articles), production (recurring unit costs under the learning curve), operations and support (O&S: crew, fuel, maintenance, spares, training over the service life), disposal (end of life: demilitarization, recycling, site remediation). For a long-lived program O&S dominates; a 20 to 30 year service life typically puts 50 to 70% of the total in O&S, with RDT&E around 10 to 20% and production 20 to 30%. The split is program specific, not a rule.
- Cost estimating relationship (CER): cost = a * x**b, a power law with driver x (mass, flight hours, thrust, quantity). The coefficients come from historical program data; a CER is valid only inside the driver range of its source data. Extrapolation is the top source of CER error.
- Learning curve: c_n = c1 * n**s with s = ln(lc) / ln(2), lc the learning curve percentage (0.80 to 0.95 for airframe production, 0.85 typical). Each doubling of the unit number drops the unit cost to lc times the previous value, so c2 = c1 * lc.
- Cumulative average unit cost: the exact discrete average of the first n unit costs, sum(c_k, k=1..n) / n. The closed form n**(s+1)/(s+1) used by parametric-cost is an approximation of the cumulative total, not the exact discrete sum.
- Present value: pv = fv / (1+i)n for a single future amount; for a uniform annual series, pv = a * (1 - (1+i)-n) / i. The discount rate i is real (inflation removed): 2 to 3% typical for government programs, higher for commercial programs.
- Inflation: escalate a then-year cost with (1+f)**years or deflate with a base-year index ratio. Real versus nominal: the subtraction r = n - f is an approximation; the exact relation is (1+r) = (1+n)/(1+f).
- Cost drivers: airframe mass, complexity and technology readiness, production rate and quantity, flight hours and cycles, fuel price, labor rates, fleet size, support concept, reliability (MTBF) and maintainability (MTTR). Weight and complexity drive RDT&E and production; flight hours and cycles drive O&S.
- Uncertainty: an early conceptual LCC point estimate carries wide bands, typically +-20 to 30%. Report a range, run sensitivity on the dominant drivers, and use Monte Carlo for confidence levels. A bare point estimate without a range is a misleading deliverable.
- FAR-25 and CS-25 are program context only; the CERs, learning curve, and discounting are common cost estimating methodology, not regulation.
Workflow
- Set the program inputs: phase costs or CER drivers, learning curve lc, discount rate i, service life years, and inflation rate f.
- Estimate each phase cost. Apply the power-law CER with cer_cost, or pass the phase estimates directly.
- Apply the learning curve with unit_cost for the Nth unit and cumulative_average_unit_cost for the average over the production run.
- Escalate then-year costs with escalated_cost and discount future year cash flows with present_value and annuity_present_value.
- Sum the phases with lcc_total; the returned dict carries the discounted O&S stream, discounted disposal, and the total LCC.
- Bound the estimate with uncertainty_range and identify the dominant cost drivers from the phase shares.
Pitfalls
- Mixing real and nominal rates: discount nominal cash flows with a nominal rate and real cash flows with a real rate; the two never mix in one calculation.
- Discounting O&S as a lump sum instead of a series: O&S is an annual stream, so it enters as an annuity, not as a single future amount.
- Extrapolating a CER beyond its data range: the power law has no built-in validity limit; the estimate silently leaves the region the coefficients were fit to.
- Treating the closed-form cumulative factor as exact: the discrete sum over units 1..n is the exact average; n**(s+1)/(s+1) is the approximation documented in parametric-cost.
- Forgetting disposal: disposal is small but mandatory in a full LCC; omitting it understates the total and misses end-of-life obligations.
- Reporting the point estimate alone: the uncertainty band and the dominant drivers matter as much as the total; a single number overstates confidence.
- Passing negative or zero inputs; the module raises ValueError instead of returning a nonsense cost.
Behavior contract (gate 3)
The phase CER, Nth unit cost, cumulative average unit cost, present value, annuity, inflation escalation, LCC rollup, and uncertainty range are exercised by the gate 3 contract test: scripts/test_life_cycle_cost.py against scripts/life_cycle_cost_logic.py (stdlib unittest, offline). Run: python3 scripts/test_life_cycle_cost.py
Compliance
- Standards referenced, not reproduced: FAR-25 is US government work (public domain) and CS-25 is a free EASA download; the CERs, learning curve, and discounting are common cost estimating methodology, summary-only per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.