Compartment Model¶

Model — instantiates Conserved Reservoir-Flux Balancing

Abstracts a system into a few well-bounded compartments linked by transfer rates, so accumulation and turnover follow from residence times instead of being watched flow by flow.

A Compartment Model lumps a messy, continuous system into a small number of well-mixed compartments connected by transfer rates. Each compartment is assumed internally uniform, so one number describes it, and the conserved quantity moves between compartments at rates that imply a characteristic residence time. The idea that makes it this mechanism: it deliberately trades spatial detail for analytic tractability. Once you accept "well-mixed boxes plus first-order transfer," turnover, half-life, and the steady-state distribution can be computed from a handful of rate constants — without tracking every parcel of the substance. It is the structural abstraction and its rate constants, distinct from a diagram that merely draws stocks and from a simulation that runs them forward.

Example¶

A drug is modeled as a two-compartment pharmacokinetic system: a central compartment (blood and well-perfused organs) and a peripheral compartment (muscle and fat), with transfer-rate constants moving drug between them and an elimination channel — clearance — carrying it out of the central compartment. The boundaries are defined by mixing speed, not anatomy: everything that equilibrates with blood quickly is lumped into "central." From just those rate constants the model yields a residence time and terminal half-life (say ≈8 hours), which tells clinicians the dosing interval that keeps plasma concentration inside the therapeutic band without the drug accumulating dose over dose. The same three-parameter structure that describes a drug in a body describes a nutrient in a lake or a cohort in an epidemic — which is the point.

How it works¶

Choose compartments by well-mixed-ness, not geography. Lump everything that equilibrates fast into one box; the compartment count is set by dynamics you must capture, not by how the system looks.
Draw each boundary so a unit sits in exactly one compartment — no double counting across boxes.
Assign first-order transfer-rate constants on the topology. Each rate encodes a residence time (residence time ≈ 1 / rate out).
Read turnover and steady state analytically from the rate matrix, rather than measuring every flow.

Tuning parameters¶

Number of compartments — one lumped box versus many. More compartments fit richer dynamics but demand more rate data and risk parameters that cannot be identified from the data you have.
Rate-law form — first-order (proportional to amount) versus saturable (Michaelis–Menten). Saturable captures a capacity limit but makes the model nonlinear.
Boundary sharpness — how cleanly compartments are separated; fuzzy boundaries reintroduce the internal mixing the model was built to abstract away.
Time-invariance — constant rate constants versus time-varying ones; constants are analytically clean but miss adaptation and regime change.

When it helps, and when it misleads¶

Its strength is yielding residence time, turnover rate, and accumulation risk from very few parameters, in a form that ports across pharmacology, ecology, and epidemiology unchanged.

It misleads when its founding fiction breaks. The well-mixed assumption^[1] fails whenever a compartment has real internal gradients — a depot that releases slowly, a subpopulation that mixes poorly — and the lumped rate then averages over structure that matters. Over-compartmentalizing produces rate constants that no data can pin down, and a fitted model is easily over-read as mechanism when several different structures fit the same curve equally well. The discipline is to justify each boundary on physical grounds and to check that the parameters are actually identifiable before trusting them.

How it implements the components¶

Compartment Model realizes the structural-abstraction side of the machinery — the geometry and kinetics, not the substance or the run:

reservoir_boundary_definition — draws each compartment's boundary so every unit belongs to exactly one, defined by mixing speed.
residence_time_or_turnover_model — the transfer-rate constants convert directly into a residence time and turnover for each compartment.
topology_map — the graph of which compartments exchange with which, plus the elimination channel out of the system.

It does not name the conserved substance or its boundary sources and sinks (conserved_quantity_definition, source_sink_register) — that's Material Flow Analysis; and it does not run the structure forward through time under scenarios (scenario_simulation_model) — that's System Dynamics Simulation, which exercises this model.

Instantiates: Conserved Reservoir-Flux Balancing — the Compartment Model is the archetype's analytic backbone: reservoirs plus transfer kinetics.
Sibling mechanisms: System Dynamics Simulation · Stock-and-Flow Diagram · Material Flow Analysis · Mass-Balance Table · Capacity Headroom Alert · Data Lineage Balance Check · Flow Gate or Valve Rule · Inventory Reconciliation Workflow · Loss-Sink Audit · Reservoir Balance Dashboard · Sankey Flow Map · Unit Conversion Crosswalk · Water or Resource Budget

Notes¶

A compartment model is a structure, not a running simulation. Its rate constants and residence times are fixed relations; feeding them to System Dynamics Simulation is what makes the stocks actually move over time. Keeping the structure separate from the run means a rate can be re-estimated from new data without rebuilding the topology.

References¶

[1] A compartment model assumes each compartment is well-mixed — uniform throughout, so a single concentration describes it. This lumped-parameter assumption is exactly what buys the analytic residence time; when internal gradients matter, the compartment must be split further or the abstraction abandoned. ↩