Advection-Diffusion or Transport Modeling¶

Transport model — instantiates Effective-Input Delivery Assurance

Predicts how much of an input reaches the target — and how much washes out or piles up along the way — by modeling its advective and diffusive transport through the delivery path.

Advection-Diffusion (or Transport) Modeling works out where an input actually goes by simulating the physics of its movement: advection (bulk flow carrying the input along) and diffusion (spreading down concentration gradients), together with the removal and buildup that happen en route. Its defining move is that it is mechanistic and predictive — it computes the concentration reaching the point of action over space and time from first principles (a velocity field, a spreading rate, a decay term), before any delivery is attempted. Where an assay measures the surviving fraction after the fact and a loss map tallies it, this model forecasts the whole profile — including exactly where the input is stripped away and where it lingers.

Example¶

A site team plans to remediate a groundwater contaminant plume by injecting a treatment reagent from a well up-gradient, about 80 m away. Before drilling more wells, they model the transport: advection at the local groundwater velocity (≈0.3 m/day), diffusion and dispersion spreading the reagent as it travels, and a first-order decay term for how fast it degrades. The model predicts that a single injection arrives at the plume core badly diluted and roughly a year late, that much of it is cleared past the plume down-gradient before it can react, and that it accumulates in a low-flow silt lens off to one side. The output reshapes the field program: staged injections and a repositioned well, chosen on a simulated profile rather than a hopeful guess about "it'll spread."

How it works¶

The model states a governing transport relationship — an advection term, a diffusion/dispersion term, and a reaction or removal term — and parameterizes the path with a velocity field, a spreading coefficient, and a decay/clearance rate. Solving it yields concentration as a function of position and time, from which it reads the amount reaching the point of action, the fraction lost to clearance before arrival, and any zone where slow clearance lets the input pile up. What distinguishes it from its siblings is that all of this is computed ahead of delivery, so the path can be redesigned before it is paid for.

Tuning parameters¶

Advection vs. diffusion balance — whether transport is flow-dominated or spreading-dominated (the Péclet regime). It decides which lever matters: more push, or more time and mixing.
Dispersivity / mixing — how much the input spreads as it travels. More spreading widens reach but drops the peak concentration at the target.
Decay / clearance rate — how fast the input is removed en route. Raising the modeled rate shrinks the predicted dose that survives to the target.
Spatial and temporal resolution — grid and timestep fineness. Finer resolves hotspots and accumulation lenses but costs compute and demands more data.
Source schedule — a single pulse versus continuous input. This reshapes both breakthrough timing and where accumulation builds.

When it helps, and when it misleads¶

Its strength is unique among the siblings: it predicts the delivery profile before delivery, and it locates loss spatially — not just that fraction was lost, but where along the path it drained away or collected. That makes it the tool for redesigning the route rather than only auditing the outcome.

Its central weakness is that a transport model is only as good as its parameters, and velocity fields, dispersivity, and decay rates are notoriously uncertain — a crisp-looking predicted plume can be confidently wrong. The classic misuse is to calibrate the model to one dataset and then extrapolate it outside that regime, or to trust the simulation over a cheap physical check. The discipline that keeps it honest is to carry the parameter uncertainty into the prediction as a band, validate the forecast against a tracer or breakthrough measurement, and treat the model as a hypothesis about the path rather than a readout of it. The Péclet number^[1] — the ratio of advective to diffusive transport — is the first sanity check on which regime you are even in.

How it implements the components¶

Transport Modeling fills the path-and-kinetics slots of the archetype — the components that describe the journey, not the endpoint verdict:

delivery_path_and_stage_map — the model is a quantitative map of the path geometry and the stages the input passes through on its way to the target.
clearance_path — the removal term represents where and how fast the input leaves the system before reaching the point of action.
accumulation_threshold — it predicts the zones where slow-clearing or repeated input builds up toward a harmful concentration.

It does not measure the surviving fraction empirically (stagewise_availability_measurement, handled by Stagewise Availability Assay) or tally proportional losses after the fact (loss_or_leakage_accounting / effective_availability_ratio, handled by Sankey Loss Map); it forecasts the profile rather than auditing the actuals.

Instantiates: Effective-Input Delivery Assurance — this model supplies the predicted delivery profile the rest of the assurance loop measures against.
Consumes: the supplied-input baseline (its source term) from Mass Balance or Sankey Loss Map.
Sibling mechanisms: Stagewise Availability Assay · Tagged Input Tracing · Availability Funnel Dashboard · Target-Site Sampling or Proxy Validation · Sankey Loss Map

Notes¶

The model predicts; it does not confirm. Its forecast should be paired with an empirical check at the target — a tracer test, a breakthrough curve, a direct assay — precisely because its most confident output (a smooth concentration field) rests on its least certain inputs (the path parameters). A validated transport model is powerful; an unvalidated one is a persuasive picture.

References¶

[1] The Péclet number is the dimensionless ratio of advective to diffusive transport rate; a high value means bulk flow dominates and a low value means spreading dominates. It tells you which physical term controls delivery — and therefore which intervention will actually move the outcome. ↩