Microfluidic, reacting, biological, and geophysical flows

A quiet stream can carry an unmixed solute

A dilute solute enters a 10 mm long channel with a prescribed axial transport speed of 1 mm/s. Its concentration is higher near one side than the other. The transverse distance is 100 μm, and the transit takes 10 s. Will the outlet concentration be nearly uniform? The answer changes when the solute diffusivity changes, even if the fluid motion stays the same.

For a supplied water-like density of 1000 kg/m³ and viscosity of 0.001 Pa·s, the width-based Reynolds number is 0.1. This compares axial inertial and viscous scales; it does not measure concentration uniformity. In the exact transport example developed below, changing diffusivity from 10⁻⁹ to 10⁻¹¹ m²/s changes the remaining relative concentration amplitude from about 0.0000414 to 0.7248. Small inertia and efficient mixing are different questions.

Microfluidic, reacting, biological and geophysical flows share conservation laws, but their useful reductions retain different processes. A laboratory channel may need molecular diffusion and surface forces; a reactor needs species conversion and heat accounting; a tissue needs uptake and boundary supply; a large rotating layer needs pressure, Coriolis acceleration and thickness changes. A name for the setting does not select its equations.

On a first reading, follow the channel comparison, the premixed reactor, the supplied tissue slab and the rotating-layer force balance. The tables give the quantitative equivalents of the figures. Sections marked “Further” provide return paths to transport equations, actuation, pulsation, heat coupling, stratification and potential vorticity. “Conservation of mass, momentum, and energy”, “Pressure, buoyancy, and fluid statics”, and “Vorticity and circulation” supply published preparation; the needed quantities are also introduced locally.

Choose the process, distance and time

Advection carries a property with fluid motion. Diffusion spreads it through molecular motion down a concentration gradient in the simple model used here. Reaction converts species; uptake is a consumption term. A concentration c is amount of a species per volume, measured here in mol/m³. A diffusive flux J measures amount crossing unit area per time, in mol/(m²·s); a reaction source R measures amount produced per volume per time, in mol/(m³·s). A negative R consumes that species.

For speed U over an axial length L, the transit scale is tₐ = L/U. Diffusion across a transverse distance w has scale tD = w²/D, where D is diffusivity in m²/s. For first-order consumption R = −kc, k has units s⁻¹ and the chemical time is tᵣ = 1/k. Shortening the diffusion distance fivefold reduces tD twenty-fivefold without changing D. The axial travel length and the transverse diffusion distance must not be interchanged.

The Péclet number Pew = Uw/D compares transport scales built from the same width w. In a long channel, the more direct residence comparison is χ = tₐ/tD = DL/(Uw²) = (L/w)/Pew. Thus Pew can exceed one while diffusion still substantially removes a particular transverse pattern before the outlet. The Damköhler number used for the ideal reactor is Da = tₐ/tᵣ = kL/U. Other geometries use other diffusion lengths or reaction times; state the definition whenever using the name.

A dimensionless ratio estimates the relative sizes of specified terms. It is not a universal switch. Boundary conditions, forcing frequency, inlet patterns and the required accuracy still decide which approximation works. The same fluid can need one balance for its velocity and another for its concentration.

Follow one concentration pattern to the outlet

To calculate rather than guess, prescribe a uniform axial transport speed U and let y run across a transverse interval 0 ≤ y ≤ w. Impose no diffusive solute flux through its walls. At the inlet set c(y,0)/c̄ = 1 + a₀ cos(πy/w), with mean c̄ = 1 mol/m³ and initial amplitude a₀ = 0.8. The inlet ranges from 1.8c̄ at y = 0 to 0.2c̄ at y = w. The concentration remains positive, and the cross-sectional mean is c̄.

For constant D, negligible axial diffusion and the prescribed U, the steady equation is U∂c/∂x = D∂²c/∂y². The outlet pattern follows exactly as c/c̄ = 1 + a(x)cos(πy/w), where a(x) = a₀ exp[−π²Dx/(Uw²)]. The cosine has zero slope at both walls, so it respects the no-flux boundaries. Taking its second derivative supplies −π²/w², which explains both the exponential decay and its length scale.

The table compares three assigned cases at the same 10 s transit. The slow solute in a 100 μm interval retains most of its inlet variation. Giving that same solute a 20 μm diffusion interval reduces the remaining amplitude to about 0.06784. The narrower case changes the assigned width while keeping U and L fixed; it does not hold a device pressure drop or throughput fixed.

This uniform-speed scalar model is not a solution for no-slip pressure-driven flow. A parabolic velocity profile gives different residence times across the channel and requires its own transport solution. The calculated amplitude is exact for the supplied cosine pattern and model, not a universal time to mix arbitrary feeds. Steady three-dimensional or time-dependent low-inertia flows can stretch concentration into thinner layers; molecular diffusion then acts over the shorter distance. Chaotic advection does not require turbulent velocity fluctuations.

Further: add species without losing total mass

The material derivative Dc/Dt = ∂c/∂t + u·∇c follows a moving fluid parcel; u is the fluid velocity. For an incompressible carrier and a dilute neutral species with constant D, its balance is Dc/Dt = D∇²c + R. The left side is storage plus advection; the right side contains diffusion and net production. Supply the initial state, inlet concentration and wall flux or concentration conditions as well as this differential equation.

A general reacting mixture uses species mass fractions Yᵢ, with ΣYᵢ = 1, density ρ and mass-average velocity u. The conservative balance is ∂(ρYᵢ)/∂t + ∇·(ρYᵢu + jᵢ) = ω̇ᵢ. Here jᵢ is species mass diffusion relative to u, in kg/(m²·s), and ω̇ᵢ is chemical mass production, in kg/(m³·s). With a mass-conserving nonnuclear reaction mechanism, Σω̇ᵢ = 0; by the definition of mass-average velocity, Σjᵢ = 0. Summing recovers total continuity ∂ρ/∂t + ∇·(ρu) = 0.

Individual species are consumed and produced while total mass is retained. The sum of molar production rates need not be zero because molecules can combine or split. Stoichiometric coefficients, molecular masses and elemental balances constrain a consistent mechanism. Assigning an independent Fick law to every concentrated-mixture species without enforcing the zero-sum mass flux can violate the chosen velocity definition. A dilute tracer law is a conditional reduction, not the entire multicomponent theory.

Momentum still balances pressure, viscous stress and body forces; the energy balance still accounts for heat, work, transported enthalpy and storage. Composition may alter density, viscosity and thermodynamic properties, while temperature alters reaction rates. If compressibility or shocks matter, use the additional development in “Compressible flow and shock waves”. If distinct phases exchange material, use the interfacial balances in “Multiphase and interfacial flows”.

A premixed reactor converts during residence

Now use a separate ideal plug-flow reactor with the same assigned U = 1 mm/s and L = 10 mm. A and B have equal molar mass and convert by A → B with supplied first-order k = 0.2 s⁻¹. The feed is already mixed: cA0 = 2 mol/m³ and cB0 = 0. The carrier density is constant, axial dispersion and wall losses are neglected, and k is held fixed. These are reactor assumptions, not consequences of a small channel.

A slice balances incoming and outgoing advected A against its consumption. At steady state, U dcA/dx = −kcA. Divide by cA and integrate from inlet to x: ln(cA/cA0) = −kx/U. Thus cA = cA0 exp(−kx/U), cB = cA0 − cA, and the converted fraction is X = 1 − exp(−Da) at the outlet. Here Da = 2, so cA ≈ 0.270671 mol/m³, cB ≈ 1.729329 mol/m³ and X ≈ 0.864665.

The residence time is 10 s and the supplied reaction time is 5 s. This explains the substantial conversion of this already mixed feed. It does not prove that two initially separated reagents contact each other. The slow-solute, 100 μm channel has a mode decay time of about 101 s, longer than both times. A fast local reaction can consume material near a contact region while leaving remote reactant poorly supplied. A cross-section-averaged rate must then account for the actual concentration field.

For a nonlinear reaction, averaging and evaluating the rate generally do not commute. For example, a local bimolecular rate proportional to cA cB depends on the mean of that product, not only the product of the separate means. A stirred reactor, a plug reactor and a diffusion-fed reaction layer therefore need different transport descriptions even when the same intrinsic chemistry is supplied.

Further: reaction changes the energy accounting

An isothermal reaction model assumes the temperature is held sufficiently constant for its supplied rate law. An adiabatic model instead excludes external heat transfer. These conditions differ. For the assigned reactor, use a separate constant-property estimate with reaction enthalpy ΔH = −100 000 J/mol, density 1000 kg/m³ and heat capacity cp = 4200 J/(kg·K). A negative ΔH denotes an exothermic conversion.

With negligible kinetic and potential changes, the estimated adiabatic temperature rise is ΔTad = (−ΔH)(cA0 − cA)/(ρcp). The numerator is released chemical enthalpy per carrier volume, in J/m³; the denominator is heat capacity per volume, in J/(m³·K). At the outlet, this gives about 0.04117 K. This small assigned temperature scale is useful to inspect, but without the temperature sensitivity of k it does not by itself establish an isothermal error bound.

Chemical energy has not been created. If a total-energy or enthalpy formulation already includes species formation energies, reaction changes composition and the energy partition consistently; adding the same reaction heat again would count it twice. A sensible-temperature formulation can instead display a heat-release source derived from species enthalpies and production rates. Keep the convention consistent with transported species enthalpy and heat conduction.

Real reacting flows can have strong thermal feedback, variable properties and stiff kinetics, meaning that some chemical times are much shorter than the resolved transport time. Ignition, extinction and a propagating flame require reaction, diffusion and thermal boundary conditions together; a single Da does not predict them. Averaged turbulent reacting flows additionally need closure for correlations and unresolved mixing. “Turbulence” develops the averaging problem; this lesson identifies the extra chemical and thermal coupling without supplying a universal combustion model.

A supplied face can leave a low concentration at the center

Biological flow also transports material to regions where it diffuses, binds or is consumed. To isolate supply versus uptake, consider a stationary homogeneous tissue or gel slab. Take its symmetric half, from center x = 0 to supplied face x = H. The face is maintained at dissolved concentration cs = 0.2 mol/m³. Inside, D = 2 × 10⁻⁹ m²/s and first-order uptake is kc with k = 0.1 s⁻¹. These supplied teaching parameters are not physiological measurements.

At steady state with no convection or binding, 0 = D d²c/dx² − kc. Symmetry gives dc/dx = 0 at the center; the face condition gives c(H) = cs. Define the penetration length ℓ = √(D/k), here about 141.421 μm, and dimensionless half-thickness φ = H/ℓ. The hyperbolic cosine is cosh z = (eᶻ + e⁻ᶻ)/2. Solving the two-boundary problem gives c(x) = cs cosh(x/ℓ)/cosh(φ). Unlike an arbitrary exponential from one face, this solution respects the zero center flux.

At H = 100 μm the center is about 0.158656 mol/m³. At H = 300 μm it is about 0.047270 mol/m³ even though the face concentration and material coefficients are unchanged. The thicker slab consumes more total solute per supplied face, but its volume-averaged uptake is a smaller fraction of the rate that would occur if every point were held at cs. More tissue does not mean proportionally more uniformly supplied tissue.

The inward flux at the supplied face is Jin = Dcs tanh(φ)/ℓ. It equals the integrated uptake ∫₀ᴴkc dx, in mol/(m²·s), for this steady half-slab. Its effectiveness η = Jin/(kcsH) = tanh(φ)/φ compares actual uptake with uniformly supplied uptake. As φ becomes small, η approaches 1; for large φ, η decreases approximately as 1/φ. First-order uptake stays positive at every positive concentration; this model does not produce a sharp depleted front or a clinical thickness limit.

Flow can maintain the face supply while diffusion crosses the tissue. A finite reservoir, membrane resistance, saturating uptake, reversible binding or interstitial advection changes the equations or boundary conditions. Dissolved concentration is also different from total amount carried by a binding protein. Those mechanisms must be measured or specified before translating this slab into an organ, culture device or treatment prediction.

Further: biological motion is not one rheological regime

A small swimmer illustrates a different biological question: can an organism move by deforming its surface? Using assigned length 10 μm, speed 20 μm/s, density 1000 kg/m³ and viscosity 0.001 Pa·s gives Re = 0.0002. The viscous momentum-diffusion time L²/ν is 0.0001 s, where ν = μ/ρ is kinematic viscosity. For an assigned 0.1 s stroke period, their ratio is 0.001. Both estimates support neglecting fluid inertia for that scale comparison.

In a Newtonian Stokes model, instantaneous pressure and viscous stresses respond linearly to the prescribed boundary motion. For a swimmer with no external force or torque, in an otherwise unforced fluid with a fixed environment, retracing the same shape sequence reverses the displacement. A reciprocal periodic stroke therefore gives no net swimming in this limit, even if one part is executed faster. Traveling flagellar waves or sequences with more than one independent shape change can avoid this cancellation, although a nonreciprocal stroke alone does not guarantee propulsion. Fluid memory, inertia or deforming surroundings require reconsidering the argument.

Blood transport raises other questions. Hematocrit is the red-cell volume fraction. The apparent viscosity measured from tube flow can depend on hematocrit, tube diameter and cell arrangement; a single constant viscosity is an approximation for a specified regime. A cell suspension can require a continuum constitutive model, resolved cells, or an averaged two-component description. “Multiphase and interfacial flows” provides the amount-versus-flux and relative-motion language; the constitutive response still needs evidence.

Pulsation can matter even when a flow remains orderly. For angular forcing frequency ω = 2πf, tube radius R and kinematic viscosity ν, the Womersley parameter α = R√(ω/ν) compares unsteady and radial viscous scales: α² = (R²/ν)/(1/ω). Small α favors rapid radial viscous adjustment relative to forcing; larger α requires an unsteady profile and phase response. This does not measure turbulence and does not establish that a vessel wall is rigid.

With assigned ρ = 1060 kg/m³, μ = 0.004 Pa·s, f = 1 Hz and speed scale 0.02 m/s, a rigid Newtonian tube of radius 1 mm has α ≈ 1.2904 and diameter-based ReD = 10.6. At radius 5 mm, α ≈ 6.4518 and ReD = 53. These are diagnostics, not calculated pulsatile velocity fields. Compliant-wall waves, non-Newtonian response and cell-scale effects need their own models; “Fluid–structure interaction and biomechanical continua” supplies the wall-motion connection.

Further: small inertia does not remove surface physics

A large surface-to-volume ratio makes wall conditions especially consequential in a small device. A liquid wall may bind solute or drive surface reactions. A meniscus introduces surface tension and wetting; droplets add moving interfaces. The no-flux scalar walls used earlier exclude those transfers deliberately. “Surface tension and capillarity” and “Multiphase and interfacial flows” retain their detailed interface models.

An electric field can drive mobile charge near a charged wall and thereby move the liquid: electroosmosis. In the thin electric-double-layer, uniform-property, weak-field idealization with negligible imposed pressure gradient, the outer velocity is represented by ueo = −εζE/μ. Here ε is permittivity, ζ is the potential at the hydrodynamic shear plane, and E is the tangential electric field. This is an effective outer slip description of a charged layer, not proof that the liquid violates its microscopic wall condition. Overlapping charge layers, surface conduction, heating and variable ζ need a coupled electrokinetic model.

For a rarefied gas, the Knudsen number Kn = λ/L compares molecular mean free path λ with the relevant geometric or gradient length L. If molecular and flow lengths are insufficiently separated, continuum constitutive laws and no-slip boundaries need scrutiny; slip or kinetic descriptions may be required. There is no channel-size rule independent of gas state and required accuracy. Gas rarefaction, liquid interfacial slip and electroosmosis are different mechanisms. A small Reynolds number alone establishes none of them.

A rotating layer can flow along a pressure slope

At large horizontal scales, rotation can compete with the horizontal pressure gradient. Use local axes x east, y north and z up, with Coriolis parameter f > 0. The Coriolis acceleration on the right side of the horizontal momentum equation is −f k̂ × u. For a northward current this acceleration points east; a pressure increase toward the east pushes west. These two accelerations can balance while the current runs along the pressure contours.

Take a separate reduced-gravity shallow-water model: an active homogeneous layer of mean thickness H = 100 m over a deep resting denser layer. The supplied effective restoring acceleration is g′ = 0.01 m/s². A layer-thickness anomaly η supplies pressure potential g′η, and hydrostatic, depth-uniform interior dynamics give Du/Dt + f k̂ × u = −g′∇η. The layer also obeys ∂h/∂t + ∇·(hu) = 0, where h = H + η is positive local thickness. Reduced gravity represents a density contrast; it is not the full gravitational acceleration.

With supplied U = 0.02 m/s, L = 20 km, f = 10⁻⁴ s⁻¹ and molecular ν = 10⁻⁶ m²/s, ReL = 4 × 10⁸ but Ro = U/(fL) = 0.01. Nonlinear acceleration scale U²/L is 2 × 10⁻⁸ m/s², one hundredth the Coriolis scale fU = 2 × 10⁻⁶ m/s². A large Reynolds number can coexist with rotation-dominated interior dynamics; it does not identify a universal turbulent state.

Geostrophic balance neglects the material acceleration at leading order: f k̂ × ug = −g′∇η. For northward vg = 0.02 m/s and no meridional thickness gradient, ηx = fvg/g′ = 0.0002, or 0.2 m per km. Over 20 km that is a 4 m change, small compared with the assigned 100 m mean depth. The westward pressure acceleration and eastward Coriolis acceleration each have magnitude 2 × 10⁻⁶ m/s². The equation fixes their relative direction; it does not give a downhill current.

The one-layer long-wave speed is c = √(g′H) = 1 m/s, so Fr = U/c = 0.02. The deformation radius LD = c/|f| is 10 km; Bu = (LD/L)² = 0.25 compares that radius with the chosen horizontal length. The depth-to-length ratio H/L is 0.005. These diagnostics support considering a shallow, slowly varying balanced interior, but forcing frequency, boundaries and the actual vertical acceleration must also be checked. Small Ro alone cannot discard rapid inertial or gravity-wave transients.

Further: conserve thickness and absolute vorticity together

Geostrophic balance is diagnostic; evolution still needs conservation and boundary conditions. In the inviscid one-layer system without mass exchange, vertical relative vorticity is ζ = ∂v/∂x − ∂u/∂y. Its sum with planetary vorticity f is absolute vorticity. Taking the horizontal curl of momentum gives D(ζ + f)/Dt = −(ζ + f)∇·u, while layer continuity gives Dh/Dt = −h∇·u. Their ratio therefore obeys Dq/Dt = 0 with q = (ζ + f)/h.

This shallow-water potential vorticity has units 1/(m·s). On an f-plane, a material column that stretches from h = 100 m to h = 120 m, beginning with ζ = 0 and f = 10⁻⁴ s⁻¹, must end with ζ = 2 × 10⁻⁵ s⁻¹. The invariant remains q = 10⁻⁶ (m·s)⁻¹. Increasing column thickness increases its absolute vorticity; horizontal convergence accompanies the stretching. The formula does not authorize changing depth while silently holding the column area or mass flux fixed.

On a β-plane, f varies with latitude, so movement also changes planetary vorticity. Conserving the appropriate potential vorticity then connects parcel motion, relative rotation and thickness; departures can support Rossby-wave dynamics. Friction, diabatic forcing and layer mass exchange supply sources rather than preserving this ideal invariant. Recovering velocity from potential vorticity requires the balance relation, depth information and boundary conditions, not merely a scalar map.

Even when molecular E is small in the interior, viscous or turbulent stress near boundaries can drive cross-pressure-contour motion and vertical pumping. A molecular scale estimate is not a calibrated eddy viscosity. A geostrophic calculation therefore does not determine a wind-forced transport or the entire ocean circulation. “Vorticity and circulation” develops the underlying curl and circulation ideas; “Turbulence” addresses why unresolved stresses need separate closure.

Further: a stratified fluid has a vertical structure

A single active layer represents only a chosen vertical structure. In a continuously stratified, locally Boussinesq fluid, use a constant reference density ρ0 in inertia while retaining small density departures in buoyancy. Let b = −gρ′/ρ0 be buoyancy in m/s², positive upward for a light departure. In a simple nonrotating reference column with z upward, N² = −(g/ρ0)dρ̄/dz is the squared buoyancy frequency. A stable column has N² > 0; a small adiabatic vertical displacement experiences a restoring acceleration. In a compressible atmosphere or deep ocean, a suitable potential-temperature or potential-density formulation replaces an unqualified in-situ density test.

Rotation supplies a second frequency. Compare the advection time with both 1/|f| and 1/N, and the vertical and horizontal lengths with one another. A stratification ratio U/(NH) uses a specified vertical scale H and is different from the one-layer Fr = U/√(g′H). Internal-wave propagation and anisotropic motion can matter even when Mach number is small. Hydrostatic balance omits vertical acceleration under its own scale conditions; it does not set vertical velocity to zero.

For slowly varying geostrophic, hydrostatic Boussinesq flow on a local f-plane, write normalized pressure π = p′/ρ0, so f k̂ × ug = −∇hπ and ∂π/∂z = b. Differentiate the horizontal relation with height to obtain f∂ug/∂z = k̂ × ∇hb. This thermal-wind relation connects horizontal buoyancy gradients to vertical shear. It gives a velocity difference across depth after integration; an independent reference velocity is still needed for absolute speed.

Heat, salt and suspended material can affect density and therefore the dynamics that transport them. Their sources, diffusivities and boundary fluxes require closure alongside momentum. Compressibility, nonhydrostatic motion, instability and turbulent mixing may defeat a particular balanced reduction. Link those mechanisms to their own models rather than treating all environmental flow as one universal shallow-water layer.

Compare models by the information they still need

The common method is to name the transported quantity, identify its sources and boundary transfers, choose lengths and times, and retain the momentum and energy terms that survive. The channel and tissue both use diffusion, but one follows a moving residence history and the other a steady two-boundary supply problem. The reactor converts species but cannot infer unmixed contact from k. The rotating layer is nearly balanced horizontally while thickness and potential vorticity can still evolve.

Before using a numerical solution, check conservation residuals, nonnegative concentrations or layer thickness, initial and boundary compatibility, spatial and time resolution, and sensitivity to the chosen transport or constitutive closure. A sharper concentration front from insufficient numerical diffusion is not automatically a more accurate reaction rate; an overdiffusive scheme can artificially mix reactants. Validation compares the actual predicted observable with suitable measurements under the modeled conditions.

The four settings are useful applications of shared mechanics, not a catalogue of interchangeable formulas. The neighboring lessons “Turbulence”, “Compressible flow and shock waves”, and “Multiphase and interfacial flows” supply their own central models. Here the additional surface, chemical, biological and planetary couplings determine which of those models can be used and which new evidence remains necessary.

Worked examples

Try it yourself

1. The two 100 μm scalar cases both have Rew = 0.1. Must their outlet concentration variations match?

   Show the answer and reasoning

   No. Their diffusivities differ by a factor of 100, while Rew contains no diffusivity.

   In the supplied transport solution the decay exponent is proportional to D. The outlet amplitudes are about 0.00004138 and 0.724814. A momentum ratio alone cannot measure scalar uniformity.

2. The fast 100 μm case has Pew = 100 and L/w = 100. Find χ = (L/w)/Pew and explain whether Pew > 1 forbids substantial transverse decay before exit.

   Show the answer and reasoning

   χ = 1, and substantial transverse decay is possible.

   The residence covers 100 width-advection times. For the supplied cosine, a/a₀ = exp(−π²χ), so the remaining amplitude is very small. A Péclet comparison must be paired with the actual travel length and boundary pattern.

3. Someone applies the premixed first-order A → B conversion of 86.47% to a different bimolecular A + B → products reactor. Its chosen residence-to-reaction-time ratio is also Da = 2, but A and B enter as separated streams with slow cross-stream diffusion. Must the same conversion follow?

   Show the answer and reasoning

   No; the assumed inlet and contact field have changed.

   The worked conversion solves a supplied first-order A → B law. The new bimolecular rate depends on local cA cB, so both the kinetics and the contact field differ. A shared numerical Da does not make these different models equivalent. Separating A from product B alone would not introduce a mixing requirement into the original first-order law.

4. A nonnuclear chemical mechanism gives Σω̇ᵢ = 0 in kg/(m³·s). Must the sum of its molar production rates also vanish?

   Show the answer and reasoning

   No. Conserved mass does not require conserved molecule count.

   Molar rates must be weighted by molecular masses to form mass rates. Molecular association or dissociation can change the number of moles while retaining total mass and elemental inventories.

5. Tripling the slab half-thickness from 100 to 300 μm raises Jin from about 1.722 to 2.748 μmol/(m²·s). Does total uptake per face triple?

   Show the answer and reasoning

   No. It increases by less than threefold, while effectiveness falls from about 0.861 to 0.458.

   The interior concentration is lower in the thicker slab, so much of the added tissue consumes less than kcs. The fixed-face first-order solution gives η = tanh(φ)/φ, not a uniform interior concentration.

6. In the Newtonian Stokes swimming limit with no external force or torque with fixed surroundings, a swimmer retraces one reciprocal shape cycle but closes faster than it opens. Is that rate change alone enough for net propulsion?

   Show the answer and reasoning

   No. The reciprocal shape sequence still cancels the displacement in that model.

   Instantaneous viscous response is linear and has no fluid-inertial memory in this limit. A different timing of the same reversed shape path is insufficient; the conditions must be revisited if inertia, complex-fluid memory or the environment changes.

7. At fixed forcing frequency and ν, multiply the tube radius by five. What happens to α and α², and has a turbulent state thereby been established?

   Show the answer and reasoning

   α increases fivefold and α² twenty-fivefold; no turbulent state has been established.

   α = R√(ω/ν) compares unsteady and radial viscous scales. It does not replace a Reynolds-number analysis, a stability calculation or observed flow evidence.

8. The rotating-layer case has ReL = 4 × 10⁸ and Ro = 0.01. Which result compares nonlinear acceleration with Coriolis acceleration, and is viscosity the leading interior balance partner in the proposed geostrophic reduction?

   Show the answer and reasoning

   Ro makes that comparison; the proposed leading interior balance is pressure against Coriolis acceleration.

   U²/L is 2 × 10⁻⁸ m/s², compared with fU = 2 × 10⁻⁶ m/s². Small molecular stress in the interior does not remove boundary stresses or validate an arbitrary turbulence closure.

9. A model uses hydrostatic vertical momentum balance. Must its vertical velocity be zero?

   Show the answer and reasoning

   No. Hydrostatic balance omits vertical acceleration at its stated order.

   Continuity, forcing and boundary conditions can still require vertical motion. Balanced evolution and boundary-driven pumping are compatible with small vertical acceleration.

10. A density survey supplies the horizontal buoyancy gradient and the geostrophic thermal-wind relation. Does that determine absolute current velocity at every depth by itself?

    Show the answer and reasoning

    No; it determines vertical shear and therefore velocity differences.

    Integrating f∂ug/∂z = k̂ × ∇hb requires a reference velocity or another constraint. An integration constant cannot be inferred from the gradient alone.

What to carry forward

• Select the transported quantity, the relevant distance and the boundary conditions before forming a ratio.
• Small fluid inertia does not guarantee scalar mixing; transverse diffusion can remain slow during an axial transit.
• Species convert while total mass is conserved. Reaction and heat must use one consistent energy convention.
• A consuming tissue can have a maintained surface concentration and a much lower center concentration; flux and integrated uptake must agree.
• A small swimmer, a pulsating vessel and a cell suspension need different mechanical and constitutive assumptions.
• Large Reynolds number can coexist with small Rossby number. Geostrophic balance is a leading pressure–rotation relation, while continuity and potential vorticity govern conditional evolution.
• Layer models, continuous stratification and thermal wind answer different vertical-structure questions; missing closures or reference conditions remain part of the result.