Corridor and Bottleneck Analytics for the Freight Network
Pull-quote: “The national freight network does not fail evenly. It fails at specific places, and most of them are visible in the data years before they make the news.”
Why this matters
Freight data and freight failures live at different resolutions. The flow records describe zone-to-zone movements: so much tonnage of this commodity from here to there by this mode. But nothing breaks zone to zone. Failures happen on segments: a bridge, an interchange, a rail junction, a lock, a port gate. Corridor and bottleneck analytics is the discipline of translating between the two, projecting millions of flows onto the physical network so that the question every infrastructure program must answer, where does capacity matter most, gets an evidence-grade answer instead of a familiar anecdote.
From flows to corridors to chokepoints
Zone-to-zone flows (FAF5, ~5.7M records)
│
▼
Network assignment ── flows projected onto the
│ designated freight network (NHFN)
▼
Corridor volumes ──── tonnage and value per segment,
│ by commodity and mode
▼
Bottleneck screening ─ volume, concentration,
│ redundancy per segment
▼
Disruption simulation ─ remove the segment, watch
the network respondThe first two steps produce the corridor picture: which segments of the national network carry how much of what. The interesting analytics start at the third step, because raw volume is a misleading bottleneck signal on its own. A busy segment with parallel alternatives is a workhorse. A moderately busy segment with no alternative is a chokepoint.
What a chokepoint looks like in the data
| Signature | What it means | Why volume alone misses it |
|---|---|---|
| High volume, low redundancy | The classic chokepoint; no viable reroute | Volume ranks it behind busier segments that have alternatives |
| High commodity concentration | One industry’s entire supply line on one segment | Total tonnage looks unremarkable |
| Mode transfer point | Port gate, rail yard, intermodal terminal | Capacity is in the transfer, not the link |
| Convergence node | Many corridors funnel through one interchange | Each corridor individually looks fine |
The screening layer is where these signatures are computed. The confirmation layer is simulation: remove or degrade a candidate segment in a network disruption run and measure what reroutes, what strands, and which commodities and regions absorb the cost. A true bottleneck announces itself in that experiment; a merely busy segment does not. Ranking segments by simulated disruption cost, rather than by traffic, is the difference between an investment list and a traffic report.
There is also a time dimension the static picture misses. Because the underlying flow data carries forecast horizons as well as history, the same screening can run against future flows, and the ranking changes: a segment comfortable today becomes the top chokepoint of the next decade as the commodities it carries grow. Emerging bottlenecks are worth more to a capital program than mature ones, since they are the ones an investment can still get ahead of.
In practice
The chain works as a whole or not at all: a unified model of the public federal freight data underneath, geospatial corridor maps for the visual layer, a network disruption engine for the confirmation runs, and anomaly detection that surfaces pattern breaks and volume spikes as they emerge, so the corridor picture is a living view rather than a report frozen at publication. Built that way, a state DOT, MPO, or port operator can rank its candidate chokepoints, stress-test each one, and export the evidence with citations attached.
Closing
Chokepoints are not where traffic is worst; they are where the network has no answer to a failure. Finding them takes the full chain: assignment to put flows onto real segments, screening that weighs concentration and redundancy alongside volume, and disruption simulation to confirm which candidates actually matter. The data to run that chain is public. Run it before the failure does.