Definition
Net operational cost of consist building decisions that optimize for in-train forces on the mainline versus the intermediate terminal switching work those decisions create, measuring whether mainline safety gains exceed terminal productivity costs.
The Measurement Gap
The Measurement Gap
AI consist builders optimize car placement for in-train forces: minimizing buff and draft forces to reduce derailment risk on the mainline. This is a safety-critical optimization. But the car sequence that minimizes in-train forces may not match the destination-block-optimal sequence, creating additional classification events at intermediate terminals. The consist builder counts force-reduction wins; nobody counts the terminal switching costs those wins generate.
This gap matters because consist building is a local optimization (minimize forces for this train) that creates externalities in a global system (intermediate terminal productivity). A consist that's safe on the mainline but requires three extra switches at an intermediate yard may be a net loss for the railroad, especially at terminals with limited classification capacity.
Decision Supported:
Is the consist builder's force optimization creating more terminal work than the mainline safety risk it eliminates? Should the optimizer's force targets be relaxed at terminals with limited classification capacity? FSTI provides the net-cost accounting to make consist-building policy decisions at the system level.
Formula
Plain Language: For each train built by the AI consist builder, compute the additional switching events at intermediate terminals caused by the car sequence chosen for force optimization (compared to a destination-block-optimal sequence). Multiply by the cost per switch. Subtract the estimated in-train force risk reduction benefit.
Notation:
FSTI = Σ (Additional_switches_i × Cost_per_switch) − Σ Risk_reduction_value_i
Variables:
| Variable | Definition | Unit | Source | Classification |
|---|---|---|---|---|
| Additional_switches_i | Extra classification events at intermediate terminals for train i compared to a destination-block-optimal build | Switches | Yard management system | DERIVED |
| Cost_per_switch | Marginal cost of a classification event: crew time, fuel, track wear, dwell time | Dollars | Operations cost model | DERIVED |
| Risk_reduction_value_i | Estimated reduction in derailment probability × cost of derailment from the force-optimized vs. block-optimal build | Dollars | Train dynamics model | DERIVED |
Sign Convention: Positive = switching costs exceed safety benefits (bad tradeoff). Negative = safety benefits exceed switching costs (good tradeoff). Zero = break-even.
Edge Cases:
- Unit trains (single origin, single destination) don't generate intermediate switching. The tradeoff only applies to manifest and mixed trains. Exclude unit trains.
- Risk reduction estimation: derailment probability is small and hard to quantify precisely. Use the railroad's train dynamics simulator for relative risk comparison, not absolute probability.
- Minimum sample: 50 manifest/mixed trains per measurement period.
Proposed formula, pending validation pilots.