Sankey Loss Map¶

Flow-visualization artifact — instantiates Effective-Input Delivery Assurance

A flow diagram whose branch widths are drawn to scale, exposing where a supplied input is lost stage by stage and what fraction survives to do useful work.

Sankey Loss Map is a to-scale flow diagram: one broad arrow enters as the supplied baseline, branches peel off at each stage as losses, and a narrowing main stream survives to the point of action as the effective fraction. Its defining move is to make proportional loss visible — because the widths are drawn in proportion to the flows, the eye lands on the fattest leak without anyone computing which stage matters most. It is not the accounting identity that reconciles the numbers (that is Mass Balance) or the trace that follows individual units (that is Tagged Input Tracing); it is the picture that turns a loss table into an at-a-glance priority.

Example¶

An energy analyst draws a Sankey Loss Map of a power system. A single broad arrow enters on the left as primary energy (100 units — the baseline). At the first divergence, a wide branch drains off as conversion loss at the generating plant (the largest single stream, lost as waste heat); smaller branches peel away for transmission and distribution losses; more is lost to end-use inefficiency; and a comparatively thin arrow survives on the right as useful energy services. Drawn to scale, the diagram settles an argument by inspection: the dominant lever is conversion efficiency, not the transmission losses people tend to debate, because the conversion branch is visibly the fattest. The map's value is that it redirects effort to the widest branch instead of the most-discussed one.

How it works¶

Every flow is drawn to scale from a single entering baseline. Each point where the stream divides is a named, sized loss branch, and the terminal arrow is the surviving effective fraction. What distinguishes it from its siblings is that the widths carry the message: a loss ledger rendered as geometry, so magnitude and proportion are read directly rather than looked up. It is a communication and prioritization artifact, not a computation — it displays flows that other mechanisms measure.

Tuning parameters¶

Stage granularity — a few big stages read clearly but hide sub-losses; many fine stages are diagnostic but clutter the picture.
Loss-vs-useful boundary — where you draw the line between productive output and waste. Moving it reframes the entire story the map tells.
Units — absolute quantities preserve magnitude; percent-of-input eases comparison across systems. The choice changes what stands out.
Snapshot vs. flow-over-time — a single period versus an animated or seasonal view that reveals when losses spike rather than only that they exist.
Branch-display threshold — how small a loss to bother rendering. Hiding tiny branches declutters but can bury a large cumulative leak made of small ones.

When it helps, and when it misleads¶

Its strength is instant prioritization: the widest loss branch is the first thing to fix, and the thinness of the surviving arrow beside the fat entering one makes the gap between nominal supply and effective input impossible to wave away.

Its weakness is that the diagram looks precise regardless of how good the underlying flow data is — a clean Sankey can lend false confidence to numbers that were largely guessed, and arbitrary stage boundaries can flatter a favored intervention or hide an inconvenient loss. The classic misuse is to draw branch widths not truly to scale, or to gerrymander the stages, so the picture argues for a conclusion the flows do not support. The discipline that keeps it honest is to source every width, keep the widths faithfully to scale, and pair the map with a reconciling identity so the flows in fact balance. The technique is the Sankey diagram^[1] proper — persuasive precisely because width reads as quantity, which is exactly why the widths must be earned.

How it implements the components¶

Sankey Loss Map fills the visualized-accounting slots — baseline, losses, and surviving fraction, rendered as geometry:

supplied_input_baseline — the entering arrow: the nominal supplied amount everything downstream is measured against.
loss_or_leakage_accounting — each branch is a named, sized loss; the diagram is the loss ledger, drawn to scale.
effective_availability_ratio — the surviving terminal arrow set against the entering one is the effective fraction, read straight off the widths.

It does not enforce that the flows actually balance (the reconciling identity, handled by Mass Balance) or trace individual tagged units through the path (audit_trace, handled by Tagged Input Tracing); it visualizes the proportions rather than reconciling or following them.

Instantiates: Effective-Input Delivery Assurance — the map is the at-a-glance picture of where supplied input is lost and how little survives to act.
Consumes: the stagewise loss figures it renders, from Stagewise Availability Assay or First-Pass Loss Audit.
Sibling mechanisms: Mass Balance · First-Pass Loss Audit · Availability Funnel Dashboard · Tagged Input Tracing · Stagewise Availability Assay

References¶

[1] A Sankey diagram is a flow diagram in which each stream's width is proportional to its flow rate. It is named after the Irish engineer Riall Sankey, who in 1898 used one to depict the energy losses of a steam engine — the earliest well-known use of the form to expose where useful input drains away. ↩