Sankey diagram¶

Represent transfers through a directed node-link diagram whose band widths are proportional to declared extensive quantities, making dominant paths, splits, mergers, and accounted losses visually comparable.

Version: v1 · 2026-08-30 · History
Domain-specific #: 2706
Origin domain: data visualization
Subdomain: quantity proportional flow diagrams
Aliases: Sankey flow diagram, Quantity-proportional flow diagram

Core Idea¶

A Sankey diagram is a directed flow representation in which the transverse width of each band is proportional to a declared extensive quantity transferred along that connection. Nodes mark sources, transformations, mergers, splits, or sinks; links preserve direction and magnitude. The width encoding is constitutive. A node-link drawing without quantity-proportional bands may be a flowchart or network diagram, but it is not a Sankey diagram in the strict sense.^[1]

The author fixes a quantity, unit, system boundary, aggregation level, and accounting period; maps relevant states or processes to nodes; and routes measured or modeled transfers as bands. A common scale converts quantity to width. Splits and mergers expose how an input is allocated, while visually dominant bands direct attention to major transfers or losses. For balanced energy or material accounts, incoming and outgoing widths can be reconciled at nodes; other applications may show categories or costs, but must state when conservation is not meaningful.^[2]

Band width encodes magnitude, not causal strength, statistical certainty, or moral importance. Layout order, curvature, color, grouping, and omitted residuals can change salience without changing the underlying quantities. A Sankey diagram is distinct from an alluvial diagram, whose ribbons commonly trace membership changes between categorical stages, and from a waterfall chart, which accumulates signed contributions along one running total. Conservation is common in physical accounts but is not a universal license to force every dataset into a balanced stock-flow model.^[3]

Structural Signature¶

Declared extensive quantity. A measurable or modeled amount such as energy, mass, money, or count fixes what width means.
Common width scale. A documented proportional rule maps quantities onto transverse band widths.
Directed links. Bands carry the quantity from a source or upstream state toward a receiver or downstream state.
Nodes. Processes, states, categories, sources, or sinks provide the connection points at which bands meet.
Split and merge structure. Branching and convergence expose allocation, recombination, or transformation pathways.
Accounting boundary. A time interval, population, facility, or system determines which transfers are inside the picture.
Residual or loss treatment. Unaccounted or dissipated amounts remain visible rather than disappearing in layout.
Visual ordering. Color, position, and grouping provide secondary organization without replacing the width scale.

What It Is Not¶

Not a generic flowchart. Flowcharts encode procedural order and decisions; their arrow thickness normally carries no amount.
Not an alluvial diagram. Alluvial ribbons usually track category membership across stages rather than a declared extensive balance.
Not a waterfall chart. Waterfalls show sequential signed contributions to a running total rather than routed node-link transfers.
Not a proof of conservation. Balanced-looking widths do not establish that measurements close physically.
Not a causal graph. Direction can describe movement or accounting without identifying intervention effects.
Not a neutral view. Aggregation, ordering, color, and omitted residuals materially shape interpretation.

Scope of Application¶

The abstraction is literal wherever practitioners can identify the same constitutive roles, apply the same boundary tests, and obtain the same kind of output. The following habitats are uses of Sankey diagram itself, not metaphors based only on resemblance.

Energy balances. Displaying fuel inputs, conversion stages, useful output, and losses under one energy unit.
Material flow analysis. Tracing material inputs, products, recycling loops, emissions, and stocks within a boundary.
Cost flow. Showing monetary allocation when the account explicitly states that width represents cost rather than matter.
Transport and migration. Depicting amounts moving among places or states without mistaking the display for a causal model.
Lifecycle communication. Compressing multi-stage inventories into major pathways while retaining a link to the underlying table.
Data quality audit. Making residuals, unit mismatches, and implausible splits visible before presentation.

Clarity¶

A clear account of Sankey diagram must preserve the recognition invariant stated in the Core Idea rather than rely on the title alone. State the encoded quantity, unit, period, boundary, data provenance, and width scale. Declare whether node balance is physical, financial, categorical, approximate, or intentionally absent. Show small residuals or document the threshold below which flows were suppressed. Preserve an accessible tabular representation because area and width comparisons are not exact readouts for every viewer. These declarations are not editorial extras: each changes what observations count, which transformations are licensed, and what conclusion can be drawn. A reader should be able to reconstruct the input, the operative rule, the output, and at least one defeater from the account without consulting an implementation or guessing an unstated convention.

Manages Complexity¶

Sankey diagram manages complexity by replacing a diffuse field of observations or possible operations with a bounded role structure: declared extensive quantity supplies a measurable or modeled amount such as energy, mass, money, or count fixes what width means.; common width scale supplies a documented proportional rule maps quantities onto transverse band widths.; directed links supplies bands carry the quantity from a source or upstream state toward a receiver or downstream state.; nodes supplies processes, states, categories, sources, or sinks provide the connection points at which bands meet.; split and merge structure supplies branching and convergence expose allocation, recombination, or transformation pathways.. The compression is useful because it localizes disagreement. One can ask whether the input was properly formed, whether a constitutive relation held, whether an alternative explanation defeats the inference, or whether the output was overinterpreted. The same compression can mislead when its discarded detail is exactly what the decision requires. A reference-grade use therefore reports both the invariant retained and the information intentionally lost.

Abstract Reasoning¶

Define the system boundary and select one commensurable extensive quantity for the primary width encoding.
Reconcile source data and units before routing any bands.
Choose nodes that preserve the distinctions needed by the analytic question.
Map every included transfer to a directed link and apply one width scale consistently.
Check splits, mergers, losses, and residuals against the source account.
Use ordering and color to aid traversal without encoding contradictory magnitudes.
Publish the underlying values and state uncertainty or modeling assumptions separately from geometry.
Test the candidate interpretation against the nearest named confusable rather than accepting a shared surface feature.
State the conclusion at the same scope as the source conditions, and retain uncertainty or nonuniqueness where the construct does not remove it.

Knowledge Transfer¶

The strict upward abstraction is Representational Modality. Sankey Diagram instantiates Representational Modality because its visual medium makes transfer magnitude legible specifically through proportional band width, while suppressing detail that the chosen node and layout scheme do not carry. Within quantity proportional flow diagrams, the full mechanism transfers literally when the same roles and boundary tests recur. Beyond that domain, only the parent-level skeleton should travel. Reusing the label Sankey diagram after removing its constitutive vocabulary would hide a change of mechanism behind an analogy. The honest transfer rule is therefore two-stage: recognize the domain-specific pattern first, then lift only the parent relation that remains invariant under a substrate change.

Examples¶

Canonical¶

An engine balance begins with 100 units of fuel energy. A band of width 100 enters the engine node; bands of widths 35, 45, and 20 leave as shaft work, exhaust heat, and cooling loss. The outgoing widths close the account and immediately show which pathway dominates. If measurement uncertainty leaves two units unreconciled, a residual band or explicit note preserves that fact instead of silently resizing the other bands.

Mapped back: input and conventions → constitutive role test → bounded output → explicit interpretation and defeater check.

Applied / In Practice¶

A regional material account routes imported and domestic material through processing, use, recycling, export, emissions, and landfill. Analysts first harmonize tonnes and the reporting year, then compare the rendered widths with the numerical ledger. A policy discussion can use the picture to locate large pathways, but it returns to the source table for exact values and does not infer causation from the left-to-right layout.

Mapped back: field observation or problem → candidate recognition → confusable and limit checks → appropriately scoped conclusion.

Structural Tensions¶

T1: Visual immediacy versus numerical precision. Band widths expose large differences but support only approximate visual comparison. Diagnostic: Can every displayed value be recovered from a table or label?
T2: Accounting closure versus honest residuals. A neat balance can tempt authors to hide measurement mismatch. Diagnostic: Does the node balance follow the data, or was the data forced to fit the drawing?
T3: Detail versus legibility. More pathways improve coverage while crossings and thin bands impede reading. Diagnostic: Which aggregation changes the conclusion, and is that threshold disclosed?
T4: Direction versus causation. An arrow can be read as causal even when it only records transfer. Diagnostic: Would an intervention on the source justify the same directional claim?
T5: Physical conservation versus metaphorical flow. Energy and mass accounts obey stronger constraints than cost or category displays. Diagnostic: Is conservation a property of the tracked quantity or only a layout convention?
T6: Autonomy versus Representational Modality. The parent explains how medium shapes expression; the Sankey accent is quantity-proportional bands and flow topology. Diagnostic: Remove the width scale and test whether the artifact remains recognizably Sankey.

Structural–Framed Character¶

The proportional geometry and accounting equations are structural once data and units are fixed; node aggregation, ordering, styling, and disclosure thresholds are framed design choices. The five framing criteria point in a consistent direction. Evaluative weight is limited to whether the defining conditions are met, not whether the outcome is desirable. Human practice matters to the extent that experts choose conventions, instruments, or reporting thresholds, but those choices do not make every verdict arbitrary. Institutional history explains the name and standard use; it does not replace the recognition rule. The operative vocabulary travels within the home field and closely adjacent subfields, while transfer farther away requires translation to the parent prime. Thus recognition remains disciplined even where interpretation is defeasible.

Structural Core vs. Domain Accent¶

What is skeletal. Sankey Diagram instantiates Representational Modality because its visual medium makes transfer magnitude legible specifically through proportional band width, while suppressing detail that the chosen node and layout scheme do not carry. This is the part that can be expressed without the candidate's specialist nouns.

What is domain-bound. The irreducible accent is a directed node-link flow picture, a common width-to-quantity scale, splits and mergers, system boundaries, and explicit treatment of losses or residuals. Remove those elements and the result is no longer Sankey diagram; it is only the parent relation or a loose analogy.

Why this does not clear the prime bar. The name does not recur with unchanged diagnostics across three independent domains. What transfers is already represented by prime:representational_modality. The candidate remains autonomous because its in-domain recognition rule, failure modes, and consequences are stable, but its vocabulary and interventions do not float free of the home substrate.

Sankey Diagram instantiates Representational Modality because its visual medium makes transfer magnitude legible specifically through proportional band width, while suppressing detail that the chosen node and layout scheme do not carry.

The prospective workspace queue contains one strict upward edge to prime:representational_modality. No live DAG mutation is authorized.

Relationships to Other Abstractions¶

Current abstraction Sankey diagram Domain-specific

Parents (1) — more general patterns this builds on

Sankey diagram is a kind of Representational Modality Prime

Sankey Diagram instantiates Representational Modality because its visual medium makes transfer magnitude legible specifically through proportional band width, while suppressing detail that the chosen node and layout scheme do not carry.

Hierarchy path (1) — routes to 1 parentless root

Sankey diagram → Representational Modality → Representation → Abstraction

Neighborhood in Abstraction Space¶

Sankey diagram sits in a sparse region of the domain-specific corpus (93^rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

Alluvial diagram. Tracks regrouping or membership across categorical axes; quantitative width may be present but physical flow accounting is not constitutive.
Flowchart. Represents procedural control and decisions rather than quantity-proportional transfer.
Network flow model. An optimization or feasibility model on a graph, not necessarily a visual encoding.
Reservoir-Flux Network. A stock-flow ontology requiring named reservoirs and conserved fluxes; a Sankey is a display that can portray broader transfers.
Waterfall chart. A sequential cumulative decomposition rather than routed bands through nodes.
Chord diagram. Uses arcs around a circular arrangement and need not preserve direction or a flow balance.

References¶

[1] Kennedy, A. B. W., and Sankey, H. R. (1898). 'The Thermal Efficiency of Steam Engines.' Minutes of the Proceedings of the Institution of Civil Engineers 134, 278–312. https://doi.org/10.1680/imotp.1898.19100 registry ↩

[2] Schmidt, M. (2008). 'The Sankey Diagram in Energy and Material Flow Management, Part I: History.' Journal of Industrial Ecology 12(1), 82–94. https://doi.org/10.1111/j.1530-9290.2008.00004.x registry ↩

[3] Schmidt, M. (2008). 'The Sankey Diagram in Energy and Material Flow Management, Part II: Methodology and Current Applications.' Journal of Industrial Ecology 12(2), 173–185. https://doi.org/10.1111/j.1530-9290.2008.00015.x registry ↩