Maintenance Data, Spares and Economic Optimization

Problem Statement: **Primary relation:** `Failure rate per 1,000 operating hours=1000×N_failures/operating_hours for a chosen event definition.` The relation should first be read in words and translated into a dimensional statement. Identify which quantities are probabilities, time or cycle measures, rates, costs, loads, engineering scores or empirical indicators. If an input is a probability, its feasible interval is from zero to one; if it is an event intensity, it must be nonnegative and accompanied by a unit such as failures per operating hour. The output should be interpreted at the same stated boundary and over the same mission or reporting period as its inputs. **Reasoning sequence.** (i) Define the asset and the function; (ii) write all inputs and their units; (iii) choose the simplest physically defensible model; (iv) perform the computation using an independent check; (v) compare the result with limiting cases; (vi) state uncertainty and model boundaries before recommending maintenance. For example, the limit of a no-failure event probability as an assumed event rate approaches zero should be consistent with ordinary experience. A result outside its physically possible range suggests either an arithmetic fault, incompatible units or a model used outside its assumptions. **Worked engineering example.** A facility records eight unscheduled functional failures over 16,000 runtime hours, or 0.5 per 1,000 runtime hours. If one site excludes protective trips, direct comparison becomes invalid. After evaluating the arithmetic, report what the result means and—equally important—what it does not mean. Distinguish a hypothetical prediction from a measured population statistic. If a numerical example uses a notional fleet or synthetic data, it must not appear in an operator's work package as a measured site result. For any policy comparison, repeat the calculation using at least two plausible ranges for its most uncertain input and report whether the choice changes. ### Audit model assumptions Every model has a region in which its assumptions make sense. Before using an equation in a recommendation, check units, boundary, independence assumptions, the way failures are counted, and whether repair changes the modeled population. ## 3. Where the model can fail Backfilling unknown failure reasons as “bearing fault” because the bearing was replaced creates confirmation bias and artificially high cause frequency. An important diagnostic practice is to seek evidence against the simplest explanation. A repair which temporarily restores output can disguise an uncorrected design, installation or operating error. If the apparent rate improves after a maintenance change, normalize against exposure and examine whether production load, weather, equipment population or reporting conventions changed in the same period. Correlation between a signal and a failure does not establish that intervening on the signal will prevent the failure.

Problem Statement: **Primary relation:** `For Poisson demand with mean m during lead time, P(N<=k)=exp(-m)Σ_(j=0)^k m^j/j!.` The relation should first be read in words and translated into a dimensional statement. Identify which quantities are probabilities, time or cycle measures, rates, costs, loads, engineering scores or empirical indicators. If an input is a probability, its feasible interval is from zero to one; if it is an event intensity, it must be nonnegative and accompanied by a unit such as failures per operating hour. The output should be interpreted at the same stated boundary and over the same mission or reporting period as its inputs. **Reasoning sequence.** (i) Define the asset and the function; (ii) write all inputs and their units; (iii) choose the simplest physically defensible model; (iv) perform the computation using an independent check; (v) compare the result with limiting cases; (vi) state uncertainty and model boundaries before recommending maintenance. For example, the limit of a no-failure event probability as an assumed event rate approaches zero should be consistent with ordinary experience. A result outside its physically possible range suggests either an arithmetic fault, incompatible units or a model used outside its assumptions. **Worked engineering example.** If expected demand during lead time is m=0.5, having one spare covers at most one demand with probability exp(-0.5)(1+0.5)=0.9098, leaving roughly 9% probability of two or more demands. After evaluating the arithmetic, report what the result means and—equally important—what it does not mean. Distinguish a hypothetical prediction from a measured population statistic. If a numerical example uses a notional fleet or synthetic data, it must not appear in an operator's work package as a measured site result. For any policy comparison, repeat the calculation using at least two plausible ranges for its most uncertain input and report whether the choice changes. ### Keep the operational objective in view A metric is useful when it changes a decision that helps deliver service. Additional data collection should therefore answer a concrete question: whether to inspect, redesign, change stocking policy, add monitoring or accept an explicitly bounded residual risk. ## 3. Where the model can fail Demand events may cluster due to a bad production batch or harsh season; the Poisson model underestimates risk when failures are correlated. An important diagnostic practice is to seek evidence against the simplest explanation. A repair which temporarily restores output can disguise an uncorrected design, installation or operating error. If the apparent rate improves after a maintenance change, normalize against exposure and examine whether production load, weather, equipment population or reporting conventions changed in the same period. Correlation between a signal and a failure does not establish that intervening on the signal will prevent the failure.

Problem Statement: **Primary relation:** `Equivalent annualized cost can compare life-cycle alternatives with equal service objectives and a consistent discount rate.` The relation should first be read in words and translated into a dimensional statement. Identify which quantities are probabilities, time or cycle measures, rates, costs, loads, engineering scores or empirical indicators. If an input is a probability, its feasible interval is from zero to one; if it is an event intensity, it must be nonnegative and accompanied by a unit such as failures per operating hour. The output should be interpreted at the same stated boundary and over the same mission or reporting period as its inputs. **Reasoning sequence.** (i) Define the asset and the function; (ii) write all inputs and their units; (iii) choose the simplest physically defensible model; (iv) perform the computation using an independent check; (v) compare the result with limiting cases; (vi) state uncertainty and model boundaries before recommending maintenance. For example, the limit of a no-failure event probability as an assumed event rate approaches zero should be consistent with ordinary experience. A result outside its physically possible range suggests either an arithmetic fault, incompatible units or a model used outside its assumptions. **Worked engineering example.** A repaired pump costs 2,000 today and is expected to incur 900 annual service cost; a replacement costs 8,000 and 300 annually. The least cost depends strongly on horizon, residual value and unplanned failure consequences. After evaluating the arithmetic, report what the result means and—equally important—what it does not mean. Distinguish a hypothetical prediction from a measured population statistic. If a numerical example uses a notional fleet or synthetic data, it must not appear in an operator's work package as a measured site result. For any policy comparison, repeat the calculation using at least two plausible ranges for its most uncertain input and report whether the choice changes. ### Make uncertainty actionable

Problem Statement: **Primary relation:** `A simple annual cost sample C=N_f C_f+N_pm C_pm+fixed cost; the event process for N_f must match exposure assumptions.` The relation should first be read in words and translated into a dimensional statement. Identify which quantities are probabilities, time or cycle measures, rates, costs, loads, engineering scores or empirical indicators. If an input is a probability, its feasible interval is from zero to one; if it is an event intensity, it must be nonnegative and accompanied by a unit such as failures per operating hour. The output should be interpreted at the same stated boundary and over the same mission or reporting period as its inputs. **Reasoning sequence.** (i) Define the asset and the function; (ii) write all inputs and their units; (iii) choose the simplest physically defensible model; (iv) perform the computation using an independent check; (v) compare the result with limiting cases; (vi) state uncertainty and model boundaries before recommending maintenance. For example, the limit of a no-failure event probability as an assumed event rate approaches zero should be consistent with ordinary experience. A result outside its physically possible range suggests either an arithmetic fault, incompatible units or a model used outside its assumptions. **Worked engineering example.** Sampling annual failures from a Poisson distribution with mean 3 and multiplying by failure consequence demonstrates cost variability. If failures share one common utility, independent Poisson counts can be misleading. After evaluating the arithmetic, report what the result means and—equally important—what it does not mean. Distinguish a hypothetical prediction from a measured population statistic. If a numerical example uses a notional fleet or synthetic data, it must not appear in an operator's work package as a measured site result. For any policy comparison, repeat the calculation using at least two plausible ranges for its most uncertain input and report whether the choice changes. ### Plan a falsification test A good explanation can be disproved. State an observation that would contradict the proposed failure mechanism, and arrange measurements so another engineer could reach an independent conclusion rather than merely agreeing with the initial diagnosis. ## 3. Where the model can fail Fitting broad uncertainty intervals because data are missing should be accompanied by a data-collection plan, not merely a more sophisticated simulator. An important diagnostic practice is to seek evidence against the simplest explanation. A repair which temporarily restores output can disguise an uncorrected design, installation or operating error. If the apparent rate improves after a maintenance change, normalize against exposure and examine whether production load, weather, equipment population or reporting conventions changed in the same period. Correlation between a signal and a failure does not establish that intervening on the signal will prevent the failure.

Problem Statement: **Primary relation:** `OEE=A×P×Q.` The relation should first be read in words and translated into a dimensional statement. Identify which quantities are probabilities, time or cycle measures, rates, costs, loads, engineering scores or empirical indicators. If an input is a probability, its feasible interval is from zero to one; if it is an event intensity, it must be nonnegative and accompanied by a unit such as failures per operating hour. The output should be interpreted at the same stated boundary and over the same mission or reporting period as its inputs. **Reasoning sequence.** (i) Define the asset and the function; (ii) write all inputs and their units; (iii) choose the simplest physically defensible model; (iv) perform the computation using an independent check; (v) compare the result with limiting cases; (vi) state uncertainty and model boundaries before recommending maintenance. For example, the limit of a no-failure event probability as an assumed event rate approaches zero should be consistent with ordinary experience. A result outside its physically possible range suggests either an arithmetic fault, incompatible units or a model used outside its assumptions. **Worked engineering example.** A production cell with availability 0.90, performance 0.95 and first-pass quality 0.98 has OEE≈0.8379. A maintenance intervention improves the availability term only if it reduces the counted losses. After evaluating the arithmetic, report what the result means and—equally important—what it does not mean. Distinguish a hypothetical prediction from a measured population statistic. If a numerical example uses a notional fleet or synthetic data, it must not appear in an operator's work package as a measured site result. For any policy comparison, repeat the calculation using at least two plausible ranges for its most uncertain input and report whether the choice changes. ### Define the denominator Write down the reference population and the interval across which the observation has meaning. An event per machine-day differs from an event per operating hour if production schedules differ. A simulation that silently uses an assumed denominator produces a neat but potentially false comparison. ## 3. Where the model can fail Counting planned shutdowns differently between departments invalidates cross-site benchmarking; quality loss may be caused by upstream process variation, not machinery. An important diagnostic practice is to seek evidence against the simplest explanation. A repair which temporarily restores output can disguise an uncorrected design, installation or operating error. If the apparent rate improves after a maintenance change, normalize against exposure and examine whether production load, weather, equipment population or reporting conventions changed in the same period. Correlation between a signal and a failure does not establish that intervening on the signal will prevent the failure.