Appendix: capacity and flow dictionary

The replication dictionary defines what enters and leaves the annual Corridor lab computation. Its purpose is to let a reader reconstruct the fictional studies and identify where an actual application would need more evidence. It is not a glossary of every logistics term. Each entry connects a quantity or state to the method, its unit and its decision role. A reader should be able to explain why a value belongs in the annual minimum rather than merely recognise its label.

Demand is the chosen annual quantity of eligible service in the fictional chain. It is not a sales forecast, production aspiration or measured market estimate. The app uses it as an upper bound on common flow. A lower demand fixture can therefore make the chain demand-limited without changing any asset. For an actual application, the analyst should define what qualifies as eligible demand, whether it is contracted or assumed and which acceptance boundary the quantity refers to.

A stage is a required serial function through which all modelled cargo passes. Its name identifies that function within the fixture. Adding an alternative route as another stage changes the meaning incorrectly because the app will require the common flow to pass through it as well. A stage should not be inserted merely to represent an interesting stakeholder. Its capacity input must correspond to a required function that can constrain the chosen serial service.

Nominal capacity is an annual rate before the fixture's explicit availability multiplier. It uses the same cargo quantity basis across stages. It should not be confused with stored material, fleet count or instantaneous equipment rate. Any derivation from those objects lies outside the current app and needs its own validated method. The dictionary therefore makes the annual convention visible instead of pretending that a unit label proves how the input was obtained.

Availability is a percentage input that the engine applies to nominal capacity to produce effective annual capacity. It represents the fictional annual usable-capacity assumption. It does not establish a probability distribution, mechanical reliability or event timing. The input definition should avoid double-counting downtime already included in the nominal annual rate. For actual use, the team should state whether the percentage includes maintenance, crew, access or other readiness conditions and preserve that basis with the scenario.

Variable cost is the assumed operating amount per common tonne attributed to a stage. The engine sums those stage inputs. It does not calculate a supplier tariff, marginal route response or changes in cost at different operating rates. Fixed cost is the assumed annual operating obligation attributed to a stage. The engine sums it and allocates it to common throughput when that throughput is positive. The fictional costs are illustrative and represent no client, transaction or market quotation.

The quantity and currency conventions make that allocation dimensionally coherent. Annual fixed amounts and annual throughput use matching million-scale units, yielding the labelled cost per tonne. The manuscript prints canonical scalar outputs rather than recomputing them in authoring code. This prevents a separate calculation from drifting away from the public app. It does not expand the model's economic scope: capital timing, financing, taxes, replacement and residual value are absent unless separately represented by a suitable method.

Undefined unit cost is an important output state. When common throughput is zero, the engine cannot divide fixed obligations by a positive delivered quantity, so it reports an unavailable unit cost. Fixed obligations remain in the fixture. This should not be displayed as free service, a finite average or a numerical placeholder that invites comparison with positive-flow cases. The reader should instead inspect the zero-service condition and the relevant continuing obligations.

Throughput is the minimum of eligible demand and all effective stage capacities. It is a calculated annual quantity within the serial convention. The output is not an observed delivery total. It does not establish how work is scheduled, whether buffers are sufficient or whether cargo is accepted at every actual interface. Those are applicability conditions that the main chapters require the practitioner to review before treating a real input slate as a service promise.

A bottleneck output lists stage names whose effective capacity equals the common flow under the engine's comparison rule. Demand limitation can produce an empty list when every stage has capacity above demand. Equal capacities can produce several names. The interpretation should preserve both states. An empty list does not mean the service has no problem; it means the represented annual capacity stages do not bind under the chosen demand. Several names imply that isolated relief may not create common-flow gain.

Stage utilisation compares common annual flow with that stage's effective annual capacity. It is unavailable for an effective capacity that is zero. Spare capacity is the difference between effective annual capacity and common flow. These are model-balance quantities, not measured shift utilisation, stock reserves or guaranteed recovery margins. Their useful role is to show which represented capacities exceed the current annual bound and to prompt review of why those margins exist.

Stage annual cost combines its fixed operating input with variable cost at common flow. Stage unit cost allocates those amounts to common positive throughput. Whole-chain unit cost combines the stage costs using the same flow. The model assumes every tonne incurs every serial stage's variable cost. A service with losses, bypasses, transformations or rework needs another quantity and cost representation. The dictionary helps a reader identify that mismatch before accepting the displayed totals.

The app's single-stage expansion tests change one nominal capacity according to its implemented method and hold availability, demand, other stages and costs fixed. Each result reports the new effective capacity, common throughput, gain and next constraint. These are deterministic sensitivity tests. They are not procurement designs and contain no investment cost. The manuscript's frozen cases also include explicitly authored interventions, whose complete inputs are retained separately from these built-in tests.

The availability stress changes each stage's availability according to the implemented app rule, with a lower bound preventing a negative capacity assumption. It then calculates stressed common flow and operating unit cost. The result is a scenario, not a likelihood-weighted forecast. Its value is to expose dependence on the annual readiness inputs. A practitioner should not infer a confidence level or treat the fixed stress as covering every relevant interruption mechanism.

The fixture files preserve complete inputs, and the output files preserve the canonical result object, including tables, methods, warnings and limitations. Bindings identify scalar input or output paths for every printed model quantity. Claims retain the illustrative basis, source fixture, sensitivity and manuscript block references. This chain allows the audit to execute the current engine, compare exact outputs and resolve each published token without a parallel numerical implementation.

The dictionary can diagnose a reuse error through the identity being transferred. A nominal annual rate copied from a supplier test needs an availability and calendar bridge. A storage volume copied into a stage-capacity field needs a turnover mechanism; changing its unit label cannot supply one. A route alternative represented as another mandatory stage changes parallel choice into serial obligation. An undefined unit cost copied as a neutral amount erases the zero-flow denominator. These are different repairs, and none requires disputing the engine's arithmetic. Applying this translation test before computation prevents a faithful calculation from silently answering a different physical question than the one the analyst intended.