Skip to main content
Sandbox Physics
Microswimming · A course with experiments

Stokes flow and the two tests for negligible inertia

A slow flow can usually follow a small swimmer without an appreciable inertial delay. Check how quickly the stroke changes as well as how fast it moves. Open Lauga and Powers, arXiv:0812.2887v1, Sections 3.1–3.3, alongside this guide. We will use their force balance, sphere mobility and point-force solution to work an example, then identify where those calculations stop describing a swimmer.

Section 3.1: identify the two accelerations

Equation (1) describes an incompressible Newtonian liquid with constant density and viscosity. Fluid velocity is measured in metres per second, pressure in pascals, density in kilograms per cubic metre and dynamic viscosity in pascal seconds. The pressure gradient and viscosity supply force per unit volume. The left side accounts for acceleration.

The partial time derivative can be large when a motion changes rapidly, even if its speed is small. The other term arises when a moving fluid parcel enters a region with a different velocity. Dropping both terms requires two comparisons; a small speed-based Reynolds number checks only the second one.

The gradient tells how a field varies across space. The Laplacian combines second spatial derivatives of velocity, and zero divergence expresses conservation of volume for this incompressible liquid. These operators let one equation describe a whole three-dimensional flow.

Velocity u\mathbf u, pressure pp, density ρ\rho and viscosity η\eta are the fields and material constants in this balance.

ρ[∂tu+(u⋅∇)u]=−∇p+η∇2u∇⋅u=0\begin{gathered}\rho\left[\partial_t\mathbf u+(\mathbf u\cdot\nabla)\mathbf u\right]\\=-\nabla p+\eta\nabla^2\mathbf u\\\nabla\cdot\mathbf u=0\end{gathered}

Check the frequency condition as well

Use a characteristic length and speed for the part of the flow being studied. Comparing advective inertia with viscosity gives the usual Reynolds number. For motion with angular frequency, comparing the local time derivative with viscosity gives the frequency Reynolds number. Both must be small to use the instantaneous Stokes approximation. Lauga and Powers state the frequency condition immediately after Equation (3).

The viscous diffusion time is density times length squared divided by viscosity. It estimates how long momentum takes to spread across that length. The frequency condition says that this time is much shorter than the inverse angular frequency. A slow but extremely rapid oscillation can fail this test; then the local time derivative must remain.

For a ten-micrometre length in water, take density one thousand kilograms per cubic metre and viscosity one millipascal second. The diffusion time is 0.0001 seconds. A stroke with angular frequency twelve radians per second has a frequency Reynolds number of 0.0012. A speed of thirty micrometres per second gives an ordinary Reynolds number of 0.0003. These illustrative inputs satisfy both tests.

Here UU is the characteristic speed, LL the length, ω\omega the angular frequency in radians per second, and tνt_\nu the diffusion time in seconds.

Re=ρULη≪1Rω=ρωL2η≪1tν=ρL2η\begin{gathered}\mathrm{Re}=\frac{\rho UL}{\eta}\ll1\\\mathcal R_\omega=\frac{\rho\omega L^2}{\eta}\ll1\\t_\nu=\frac{\rho L^2}{\eta}\end{gathered}

Equation (3): instantaneous force balance

With both inertial contributions negligible, pressure and viscosity balance at each instant. This approximation omits acceleration from the force balance; a prescribed boundary can still change speed. A stop does not physically propagate through real water in zero time. The approximation describes the flow after the short inertial adjustment at the scales just checked.

For fixed geometry, the Newtonian Stokes equations are linear. Multiply all imposed velocities by the same factor and the velocity, pressure difference and viscous force scale by that factor. Reversing all imposed motions and forces reverses the solution at that same geometry. Shape changes still matter: a swimmer’s displacement over a stroke need not be linear in its stroke amplitude.

Here “no memory” refers to this quasi-steady Newtonian force balance. Polymer stress, an evolving concentration field or a changing boundary shape can retain a history even when inertia is small. A viscosity value alone does not establish that a real liquid meets the Newtonian assumption. Density remains necessary for checking the approximation, even though it has dropped out of the reduced equation.

−∇p+η∇2u=0∇⋅u=0\begin{gathered}-\nabla p+\eta\nabla^2\mathbf u=0\\\nabla\cdot\mathbf u=0\end{gathered}

Worked example

Section 3.2: tow one sphere through water

Consider an isolated, rigid, no-slip sphere in an unbounded Newtonian liquid at rest far away. Its radius is one micrometre and its imposed speed is thirty micrometres per second. Use the water viscosity above. The force in Stokes’ law is the magnitude of the applied towing force; the fluid drag on the sphere has the opposite direction.

Equations (4)–(6) give the drag, mobility and thermal diffusion. The drag is 0.565 piconewtons and the supplied power is 17.0 attowatts. An attowatt is a billionth of a billionth of a watt. All this mechanical power becomes heat in the steady towing problem.

For diffusion, choose an absolute temperature of 293 kelvin. Multiplying the mobility by the thermal energy gives a diffusion coefficient of 0.215 square micrometres per second. These are independently calculated teaching values, not a measured microorganism’s speed or power. Walls, other spheres and a different liquid would change the mobility.

The symbols are force magnitude FF, radius aa, speed UU, power PP, diffusion coefficient DD, absolute temperature TT and Boltzmann constant kBk_{\mathrm B} in joules per kelvin.

kBT≃4.05×10−21 JF=6πηaU≃0.565 pNP=FU≃17.0 aWD=kBT6πηa≃0.215 μm2s−1\begin{gathered}k_{\mathrm B}T\simeq4.05\times10^{-21}\,\mathrm J\\F=6\pi\eta aU\simeq0.565\,\mathrm{pN}\\P=FU\simeq17.0\,\mathrm{aW}\\D=\frac{k_{\mathrm B}T}{6\pi\eta a}\\\simeq0.215\,\mu\mathrm{m}^2\mathrm{s}^{-1}\end{gathered}

Worked example

Section 3.3: distinguish towing from swimming

Equation (7) gives the Stokeslet, the flow produced by a point force acting on the fluid. Distance is measured from that point; the unit radial vector gives its direction, and the identity tensor leaves a vector unchanged. The formula predicts an inverse-distance velocity field. Along the force direction it gives twice the speed found at the same distance perpendicular to the force.

Insert the towing force from our example. Ten micrometres ahead of the sphere centre, the Stokeslet predicts 4.5 micrometres per second. The exact translating-sphere solution also has an inverse-cube correction; here that lowers the axial speed to 4.485 micrometres per second. The point-force result is a far-field approximation, increasingly accurate as distance exceeds the sphere radius.

A freely swimming organism has no external towing force in this idealization. The forces it applies to the fluid sum to zero, so their leading Stokeslets cancel. A force dipole often supplies the next term, with inverse-square decay; symmetries can cancel that term too. The inverse-distance example therefore belongs to the externally towed sphere. It is not the universal far field of a swimmer.

In the three-sphere approximation, point-force interactions connect separate beads. The approximation requires bead radii small compared with their separations. Adding the separate point forces is useful locally, but their total-force cancellation must still be respected when asking about the distant flow.

Position r\mathbf r is measured from the point force, rr is its magnitude, r^\hat{\mathbf r} its unit vector, and I\mathbf I the identity tensor. The force F\mathbf F acts on the fluid; in steady towing it equals the external force on the sphere.

u(r)=18πηr⋅(I+r^r^)⋅Fu∥=F4πηr=3a2rU\begin{aligned}\mathbf u(\mathbf r)={}&\frac{1}{8\pi\eta r}\\&\cdot(\mathbf I+\hat{\mathbf r}\hat{\mathbf r})\cdot\mathbf F\\u_\parallel={}&\frac{F}{4\pi\eta r}=\frac{3a}{2r}U\end{aligned}

Thermal motion changes what a trajectory means

The diffusion coefficient gives a root-mean-square displacement of about 0.46 micrometres along one axis in half a second. Root mean square measures the typical size of a random displacement; it does not specify a preferred direction. In three dimensions the mean squared displacement is three times the one-axis value.

A swimmer’s deterministic path and a bead’s Brownian path are different calculations. A three-bead swimmer needs its collective mobility to predict translational diffusion, and rotational diffusion can change its swimming direction. Comparing a mean stroke with the diffusion of one isolated bead gives a scale estimate, not a universal number of cycles after which swimming becomes visible.

When reading another calculation, identify the liquid and boundary conditions, check both inertial ratios, decide whether there is a net external force, and ask whether the path is a deterministic mean or includes thermal fluctuations. Those four choices determine which of the sphere results can be reused.

Here Δx\Delta x is displacement along one axis and tt is elapsed time; angle brackets denote an ensemble average.

⟨(Δx)2⟩=2Dt\sqrt{\langle(\Delta x)^2\rangle}=\sqrt{2Dt}

An experiment to think through

Predict, change, explain

  1. For the prescribed four-stroke loop in the three-sphere Lab, predict how tripling viscosity changes distance per cycle and dissipated energy.
  2. Keep the drawn loop and arm speed fixed, then change Fluid viscosity. Restore it and change Arm speed. In the rheometer, compare two speeds using the Newtonian reference.
  3. For a prescribed stroke in quasi-steady Stokes flow, viscosity rescales the required forces, and speed rescales force and cycle duration. Explain why displacement per completed loop stays fixed. The prediction assumes an imposed shape sequence. With a fixed motor force, the stroke itself can change.

The three-sphere Lab couples three equal spheres through point-force Stokes flow in unbounded water. It leaves out walls, neighbours and thermal noise. The rheometer Lab models steady flow of Newtonian and shear-thinning liquids in a sheared gap.

Try explaining it yourself

1. Double the sphere radius at the same speed and temperature. What happens to drag, power and diffusion?

Read an explained answer

Drag and power double to 1.13 piconewtons and 33.9 attowatts. Mobility and diffusion halve, giving 0.107 square micrometres per second. Recheck the inertial limits because the characteristic size has increased.

2. Keep the ten-micrometre size and thirty-micrometre-per-second speed, but raise angular frequency to one hundred thousand radians per second. Is the steady Stokes approximation still justified?

Read an explained answer

The ordinary Reynolds number remains 0.0003, but the frequency Reynolds number becomes 10. Local inertia cannot be neglected. A small speed-based Reynolds number alone was insufficient.

3. Can you use the towed sphere’s inverse-distance far field for every freely swimming organism?

Read an explained answer

No. With no external force, the forces on the fluid sum to zero and the leading Stokeslets cancel. A force dipole often remains; additional symmetries can make the decay faster still.

4. The rheometer torque doubles when its speed doubles. Does that prove the sample has no elastic memory?

Read an explained answer

It supports a constant apparent viscosity at those steady speeds, after calibration and geometry are accounted for. A viscoelastic liquid can have the same steady shear relation. A relaxation or oscillation experiment is needed to test its time response.

Sources to read alongside

  1. Purcell, Life at low Reynolds number, American Journal of Physics 45, 3 (1977)

    Pages 3 to 6 were read: the Reynolds-number estimates of Figure 3, the coasting example of Figure 4, the scallop theorem of Figure 6, the two-hinge swimmer of Figure 7 and the flexible oar and corkscrew of Figure 9. Pages 7 to 11 were not read for this course.

  2. Lauga & Powers, The hydrodynamics of swimming microorganisms, arXiv:0812.2887v1 (2008); Reports on Progress in Physics 72, 096601 (2009)

    Preprint version read. Sections 3 and 4 supply the Stokes equations, the point-force flow and the proofs of the scallop theorem; Sections 5 to 7 cover resistive-force theory, flexible filaments, walls and synchronisation.

Keep reading

What this page links to directly

See this series on the content map →

Microswimming · Reading path

Swimming without inertia, from Purcell’s scallop to pairs of swimmers

  1. ArticleA microbe stops the moment it stops pushing
  2. You are hereStokes flow and the two tests for negligible inertia
  3. ArticleOpening and closing a shell gets you nowhere
  4. ArticleThree beads, two arms and a mobility matrix
  5. ArticleDistance per stroke is an area in shape space
  6. ArticleA wall slows each sphere and turns the swimmer
  7. ArticleFlexible tails and turning corkscrews escape the scallop theorem
  8. ArticleTwo swimmers feel each other through the timing of their strokes
  9. ArticleReading Najafi and Golestanian’s three-sphere paper
1 more group

Other articles that use the same experiments

Each experiment from this article that another series also uses, followed by the articles there.

  1. LabRotational Rheometer & Couette–Poiseuille Flow
  2. ArticleRotational rheometers need both torque and speed
  3. ArticleStress relaxation and creep in a Maxwell liquid
Something broken, confusing, or missing on this page?Tell us
What kind of feedback?

Sent with this page’s address, its Lab ID, the site language and your device type. Please leave out personal details. Feedback is kept for up to 12 months.Privacy policy