Non-Newtonian, Microfluidic, and Biofluid Systems

Introduction

Fluid mechanics traditionally begins with Newton’s law of viscosity, in which shear stress is proportional to the velocity gradient. Many technologically important fluids do not obey this linear relation. Polymer solutions, molten plastics, paints, slurries, blood, synovial fluid, and biological suspensions exhibit viscosity that depends on shear rate, time, deformation history, temperature, concentration, or microstructure. At small length scales, additional effects become important: surface forces dominate gravity, diffusion competes with advection, and the classical assumptions of continuum fluid mechanics require careful examination.

This chapter develops a unified framework for three closely related areas. Non-Newtonian mechanics describes constitutive behavior beyond constant viscosity. Microfluidics examines flows in channels typically ranging from micrometres to millimetres. Biofluid mechanics applies fluid principles to physiological and biological systems, where fluids are often non-Newtonian, multiphase, deformable, and actively interacting with living tissue.

The central engineering challenge is not simply to select a more complicated equation. It is to identify the dominant physical mechanisms, select a constitutive model supported by data, establish the relevant dimensionless groups, and validate predictions against the operating regime.

Learning objectives

After studying this chapter, the reader should be able to:

1. Distinguish generalized-Newtonian, viscoelastic, thixotropic, and yield-stress behavior.
2. Formulate and interpret constitutive equations for common non-Newtonian fluids.
3. Analyze pressure-driven flow in a circular tube and identify the effect of rheology on the velocity and flow-rate relations.
4. Use Reynolds, capillary, Péclet, Deborah, Weissenberg, and yield-number concepts appropriately.
5. Explain the principal physical mechanisms governing microfluidic flow and transport.
6. Evaluate important features of blood flow and other biofluid systems.
7. Recognize limitations of idealized models and design appropriate experiments or simulations.

1. Continuum foundations and constitutive description

The conservation laws remain applicable to non-Newtonian fluids. For a single-phase fluid of density ρ and velocity vector u, mass conservation is

For an incompressible fluid of constant density, this becomes ∇·u = 0. Momentum conservation is

where p is pressure, τ is the extra-stress tensor, and b is body force per unit mass. The equation is not closed until a constitutive relation specifies τ in terms of deformation, temperature, concentration, and possibly the history of the material.

The rate-of-deformation tensor is

For a Newtonian fluid, the deviatoric stress is τ = 2μD, with μ the dynamic viscosity. In a non-Newtonian fluid, the apparent viscosity may depend on the magnitude of D, time, or additional internal variables. A useful scalar measure of deformation is the second invariant of D; in simple shear it reduces to the shear rate γ̇.

1.1 Generalized-Newtonian fluids

A generalized-Newtonian model retains an instantaneous relation between stress and deformation:

The apparent viscosity μ_app is a function of shear rate but not of deformation history. This class includes shear-thinning, shear-thickening, and yield-stress models in their simplest forms.

The power-law model is

where K is the consistency index, with SI units Pa·sⁿ, and n is the flow-behavior index. The corresponding apparent viscosity is

For n < 1 the fluid is shear-thinning; for n = 1 it is Newtonian with viscosity K; and for n > 1 it is shear-thickening. The power-law model is convenient over a limited shear-rate interval but predicts unphysical behavior as γ̇ approaches zero or becomes extremely large.

The Carreau model provides finite limiting viscosities:

Here μ₀ and μ∞ are the zero- and infinite-shear viscosities, λ is a time constant, and n controls the transition. The model is useful for polymeric liquids and blood-like fluids when a smooth transition between viscosity plateaus is observed.

A common yield-stress representation is the Herschel–Bulkley model:

where τ_y is the yield stress. Material that is not yielded is ideally treated as a rigid plug with γ̇ = 0 when τ < τ_y. The Bingham model is the special case n = 1. Real materials may creep below the nominal yield stress, so yield stress should be interpreted as an operational parameter obtained from a specified test protocol.

1.2 Viscoelastic fluids

Viscoelastic fluids combine viscous dissipation with elastic storage. Their response depends on time and deformation history. A simple differential model is the Oldroyd-B model, which represents a Newtonian solvent and an idealized elastic polymer contribution. In compact tensor form, the polymer stress τ_p satisfies

where λ is the polymer relaxation time, η_p is polymer viscosity, and τ_p^▽ is the upper-convected derivative. The upper-convected derivative accounts for material stretching and rotation and is essential in flows with strong extension or curved streamlines. The model is analytically useful but can predict excessive extensional stresses and does not represent finite polymer extensibility.

The Deborah number compares material relaxation time with process time:

The Weissenberg number compares relaxation time with a characteristic deformation rate:

Large De or Wi indicates that elastic effects are likely to be important. The two numbers are related but not identical: De emphasizes observation time, whereas Wi emphasizes deformation rate.

1.3 Time-dependent and thixotropic behavior

In thixotropic materials, viscosity decreases during sustained shearing and recovers during rest. Rheopexy is the opposite trend. Such behavior is described by a structural parameter, often denoted ξ, that evolves according to a kinetic equation such as

The exact terms must be established experimentally; there is no universal thixotropy equation. Consequently, a steady flow curve alone is insufficient for predicting start-up, shut-down, hysteresis, or cyclic processing.

2. Pipe flow of non-Newtonian fluids

Consider steady, fully developed, axisymmetric flow in a circular tube of radius R and length L. Assume an incompressible fluid, negligible entrance effects, no slip at the wall, and a pressure drop Δp = p_in − p_out. A momentum balance on a cylindrical fluid element gives

The wall shear stress is therefore

This result follows from momentum conservation and is independent of the constitutive law. The constitutive law determines the velocity profile and flow rate.

For a power-law fluid, τ = K(−du/dr)ⁿ in the region where velocity decreases toward the wall. Hence

Integrating from r to R and applying u(R) = 0 gives

The volumetric flow rate is

For n = 1 and K = μ, this reduces to the Hagen–Poiseuille equation:

The power-law velocity profile is flatter for n < 1 and more sharply peaked for n > 1 than the Newtonian profile at comparable conditions. The mean velocity is Ū = Q/(πR²). The wall shear rate is not generally equal to the mean shear rate. For a power-law fluid, the Rabinowitsch–Mooney correction gives the wall shear rate in terms of the flow-rate gradient:

This expression applies to the ideal power-law pipe-flow model and should not be transferred indiscriminately to arbitrary constitutive laws.

Worked example: pressure-driven power-law flow

A tube has R = 1.0 mm, L = 0.20 m, and pressure drop Δp = 20 kPa. A fluid follows the power-law model with K = 0.50 Pa·sⁿ and n = 0.60. The predicted flow rate is

The pressure factor is 100,000, raised to 1.6667, and the radius exponent is 4.6667. Evaluation gives approximately

or approximately 10.8 mL/s. This result is an idealized estimate. It assumes the power-law range includes the wall shear rates and ignores entrance losses, compressibility, slip, temperature variation, and possible elastic effects.

3. Dimensionless analysis and scaling

Dimensionless groups identify competing mechanisms and help transfer results between systems.

In non-Newtonian flow, viscosity is not known independently of the flow field. Therefore, a Reynolds number based on an apparent viscosity is a local or operating-point measure, not a universal material constant. Laminar–turbulent transition is also affected by rheology, pipe roughness, elasticity, and disturbances.

4. Microfluidic systems

Microfluidics concerns the controlled manipulation of small fluid volumes in channels, wells, droplets, or porous structures. At small scale, the surface-to-volume ratio increases as the characteristic length decreases. Pressure-driven viscous resistance often dominates inertia, while surface tension and wettability control interfaces.

4.1 Low-Reynolds-number flow

For a channel of characteristic dimension L and velocity U, Re = ρUL/μ is commonly much less than unity. The momentum equation then approaches the Stokes limit:

The flow is reversible in the ideal creeping-flow limit: reversing the pressure gradient reverses the motion without inertial memory. This reversibility makes passive mixing difficult. Mixing is therefore achieved using diffusion, chaotic advection generated by geometry, secondary flows, electrokinetic effects, or active components.

For a rectangular channel, hydraulic resistance depends strongly on aspect ratio. For a very wide slit of gap h and width w, with w much greater than h,

The cubic dependence on gap makes fabrication tolerance and fouling especially important. This formula assumes fully developed Newtonian flow, no slip, and negligible entrance effects.

4.2 Diffusion, interfaces, and electrokinetics

The species transport equation for concentration c is

where D_m is molecular diffusivity and S represents sources or sinks. In a straight channel, the ratio of advection to diffusion is measured by Pe. A large Pe means that fluid elements are transported rapidly compared with molecular homogenization, although transverse diffusion may still become important over a sufficiently long channel.

The capillary number compares viscous stress with interfacial tension. Small Ca favors nearly spherical droplets and strong surface-tension control; larger Ca allows deformation and breakup. Wettability is commonly characterized by a static or dynamic contact angle, but contact-angle hysteresis and surface heterogeneity can dominate practical behavior.

Electro-osmotic flow arises when an applied electric field acts on the electrical double layer near a charged wall. Under ideal assumptions of thin double layers, constant properties, and negligible pressure gradient, the Helmholtz–Smoluchowski velocity is

where ε is permittivity, ζ is zeta potential, E is electric-field strength, and μ is viscosity. The sign depends on the conventions for ζ and E. Real devices may exhibit Joule heating, electrochemical reactions, nonuniform surface charge, and electroviscous effects.

4.3 Slip and rarefaction

The classical no-slip condition is usually reliable for liquid microflows with hydrophilic solid surfaces, but apparent slip may arise from hydrophobic coatings, gas layers, polymer depletion, or measurement artifacts. A Navier slip condition is

where b is slip length and n is the wall-normal direction. In gas microflows, the Knudsen number Kn = λ_mfp/L compares molecular mean free path with channel size. When Kn is not small, continuum no-slip models require rarefaction corrections; at sufficiently high Kn, kinetic descriptions are necessary.

5. Biofluid mechanics

Biofluid systems differ from engineered pipes because geometry is compliant, branching, pulsatile, heterogeneous, and coupled to transport, biochemistry, and active biological regulation. Examples include blood circulation, respiratory airflow, cerebrospinal-fluid motion, urinary flow, mucus transport, and flows in cell-scale channels.

5.1 Blood as a complex fluid

Whole blood is a suspension of red cells, white cells, and platelets in plasma. Plasma is approximately Newtonian over many physiological conditions, whereas whole blood displays shear-thinning behavior, viscoelasticity, aggregation at low shear, and cell migration. The apparent viscosity depends on hematocrit, temperature, plasma composition, vessel diameter, and shear history.

At high shear rates, red-cell aggregates disperse and blood may approach a lower-viscosity regime. At low shear, rouleaux formation increases resistance. In small vessels, the Fåhræus effect causes the tube hematocrit to differ from the discharge hematocrit, while the Fåhræus–Lindqvist effect describes the reduction of apparent blood viscosity in vessels over a particular small-diameter range. These effects are not adequately represented by inserting a single constant blood viscosity into a macroscopic pipe formula.

The wall shear stress is clinically and mechanically important. For a Newtonian approximation in a circular vessel,

This relation is a special case. In blood-flow studies, the velocity profile, pulsatility, non-Newtonian constitutive law, and vessel compliance must be considered before interpreting wall shear stress.

5.2 Pulsatile flow and compliant vessels

Arterial flow is unsteady. The Womersley number compares the radius with the penetration depth of oscillatory viscous effects. Low α corresponds to a profile that adjusts nearly quasisteadily; high α produces a flatter core and stronger phase differences between pressure gradient and flow.

Vessel walls deform under pressure. A simplified thin-wall relation for a cylindrical vessel is the Laplace hoop-stress estimate:

where σθ is circumferential wall stress, p is transmural pressure, R is vessel radius, and t is wall thickness. It assumes a thin wall, uniform pressure, and neglects axial constraints, residual stress, anisotropy, and active muscle tone. Biological vessels are not passive isotropic tubes; collagen, elastin, smooth muscle, and remodeling produce nonlinear and history-dependent mechanics.

5.3 Respiratory and mucus flows

Airflow in large airways can be treated approximately as a Newtonian gas, but transitional and turbulent effects may occur near bifurcations and constrictions. Mucus is commonly modeled as a viscoelastic or yield-stress material. Its transport depends on ciliary beating, airway geometry, surface hydration, and interaction with an underlying periciliary liquid layer. A purely steady pressure-driven model cannot capture active ciliary transport.

6. Numerical and experimental methods

Computational fluid dynamics for non-Newtonian systems requires more than replacing viscosity in a Newtonian solver. For generalized-Newtonian fluids, viscosity is updated from the local shear-rate invariant, and under-relaxation may be required because strong shear-thinning creates steep viscosity gradients. Yield-stress models can cause numerical difficulties near the yielded–unyielded interface; regularization or augmented-Lagrangian methods are often used, but each introduces modeling choices.

Viscoelastic simulations must transport stress and maintain stability at high Wi. The high-Weissenberg-number problem arises from stress growth, sharp boundary layers, and numerical error. Log-conformation and related stabilized formulations improve robustness but do not eliminate the need for mesh, time-step, and constitutive-model verification.

Microfluidic experiments commonly use pressure controllers, syringe pumps, particle tracking, fluorescence imaging, micro-particle image velocimetry, and impedance or optical sensing. Calibration must account for compliance in tubing, evaporation, temperature, channel fouling, surface aging, and uncertainty in dimensions. In biofluid experiments, anticoagulants, cell viability, donor variability, and physiological temperature are important sources of variation.

Parameter identification should use experiments that excite the relevant regime. Fitting a power-law curve to steady shear data cannot determine elastic relaxation time. Oscillatory rheology, creep, stress relaxation, extensional tests, and start-up flow may be needed to distinguish competing models.

7. Illustrative case study: microfluidic blood-plasma separation

Background

Consider an illustrative diagnostic cartridge intended to separate plasma from whole blood before optical analysis. The device contains a shallow microchannel and a downstream sensing chamber. It must operate with a small sample volume, avoid hemolysis, and produce a repeatable plasma flow.

Problem

A prototype designed using a constant plasma viscosity shows variable throughput. At high hematocrit, the pressure drop increases, cells accumulate near the entrance, and the optical signal fluctuates. A Newtonian simulation predicts adequate flow but does not reproduce the observed resistance.

Analysis

The system has low channel Reynolds number, so inertia is unlikely to be the dominant cause. The relevant issues are suspension rheology, cell migration, near-wall cell depletion, entrance contraction, and possible fouling. The capillary number is also important at the plasma–air interface if the cartridge is not fully primed. Measurements should include pressure–flow curves at controlled hematocrit, temperature, and anticoagulant concentration, together with microscopy of the entrance region.

A practical first model treats the suspension with a shear-dependent viscosity calibrated over the channel shear-rate range. A more advanced model uses a two-phase or suspension-balance description with a cell concentration field. If the channel gap is comparable with cell dimensions, continuum bulk rheology alone is insufficient.

Approach and solution

The redesigned cartridge uses a widened entrance, a capillary barrier to control priming, and a parallelized separation region to reduce local flux. The model is calibrated using measured pressure drop rather than assumed viscosity. Operating limits are defined by pressure, estimated cell stress, and acceptable hemolysis markers. Quality-control tests check flow resistance and plasma yield for each production batch.

Outcome

The redesign improves repeatability because it addresses the coupled effects of rheology, geometry, interfaces, and biological variability. The key engineering decision is not the selection of the most elaborate model, but the use of a model whose assumptions match the channel scale and operating range.

Learning points

1. Blood cannot always be represented by a single constant viscosity.
2. Entrance regions may dominate the pressure loss in short microchannels.
3. Device performance requires both fluid-mechanical and biological quality metrics.
4. Calibration and uncertainty analysis are part of model validation, not optional additions.

8. Limitations, ethics, and contemporary developments

The principal limitation of non-Newtonian modeling is constitutive uncertainty. Different models may fit the same steady flow curve while predicting different transient or extensional behavior. Geometric uncertainty is also significant in microfluidics, where a small change in channel height can produce a large change in resistance.

Biofluid research requires ethical handling of human and animal materials, informed consent where applicable, biosafety procedures, privacy protection for clinical data, and transparent reporting of donor selection and exclusion criteria. A device that performs well for one hematocrit range or population may not generalize to another. Clinical or regulatory claims must be supported by validated protocols rather than simulation alone.

Contemporary research includes organ-on-chip platforms, additive manufacture of microchannels, digital microfluidics, acoustofluidics, inertial and viscoelastic cell sorting, machine-learning-assisted constitutive identification, and patient-specific cardiovascular simulation. Artificial intelligence can accelerate parameter estimation or image segmentation, but it does not replace dimensional analysis, conservation laws, uncertainty quantification, or independent validation. Training data may encode experimental bias, and extrapolation outside the training regime is particularly dangerous for nonlinear fluids.

Best practice is therefore to: define the operating range; measure rheology under relevant temperature and shear history; report geometry and boundary conditions precisely; use dimensionless analysis before simulation; verify numerical convergence; validate against independent data; quantify uncertainty; and distinguish empirical correlation from mechanistic prediction.

Chapter Summary

Non-Newtonian fluids require constitutive descriptions beyond constant viscosity. Generalized-Newtonian models describe shear-dependent viscosity, while viscoelastic models account for stress storage and relaxation. Yield-stress and thixotropic materials introduce unyielded regions and structural history. For pressure-driven pipe flow, momentum conservation determines the stress distribution, whereas the constitutive law determines velocity and flow rate.

Microfluidic systems are typically dominated by viscous, capillary, diffusive, and electrokinetic effects rather than inertia. Reynolds, capillary, Péclet, Deborah, Weissenberg, Bingham, Knudsen, and Womersley numbers provide physically meaningful scaling. Biofluid systems add pulsatility, deformable boundaries, cell suspensions, biological variability, and ethical constraints. Reliable engineering practice combines appropriate modeling, targeted rheological measurement, numerical verification, experimental validation, and uncertainty analysis.

Key Terms

Non-Newtonian fluid: A fluid whose stress–deformation relation is not characterized by a constant viscosity.

Apparent viscosity: The ratio of shear stress to shear rate at a specified operating condition.

Power-law model: A constitutive model with shear stress proportional to shear rate raised to a flow-behavior index.

Shear thinning: Decrease of apparent viscosity as shear rate increases.

Shear thickening: Increase of apparent viscosity as shear rate increases.

Yield stress: Stress threshold associated with the onset of sustained flow in an idealized model.

Herschel–Bulkley model: Yield-stress power-law constitutive relation.

Viscoelasticity: Combined viscous dissipation and elastic stress storage.

Relaxation time: Characteristic time over which elastic stress decays.

Thixotropy: Time-dependent decrease of viscosity under sustained shearing with possible recovery at rest.

Generalized-Newtonian fluid: Fluid with instantaneous shear-dependent viscosity and no explicit memory.

Creeping flow: Flow in which inertia is negligible compared with viscous forces.

Capillary number: Ratio of viscous stress to interfacial-tension stress.

Péclet number: Ratio of advective to diffusive transport.

Womersley number: Measure of unsteady inertia relative to viscous diffusion in oscillatory flow.

Electro-osmosis: Fluid motion induced by an electric field acting on charged interfacial layers.

Fåhræus effect: Difference between tube and discharge hematocrit in small-vessel blood flow.

Wall shear stress: Tangential fluid stress acting at a wall.

Constitutive equation: Material relation that closes the conservation equations.

High-Weissenberg-number problem: Numerical difficulty associated with simulating strongly elastic flows.

Review Questions

Short Answer Questions

1. What distinguishes a generalized-Newtonian fluid from a viscoelastic fluid?
2. Define the capillary number and explain its physical significance.
3. Why is the stress distribution in fully developed circular pipe flow independent of the constitutive equation?
4. State two limitations of the power-law model.
5. What physical effects make blood flow different from flow of a simple Newtonian liquid?

Long Answer Questions

1. Derive the velocity profile and flow-rate relation for a power-law fluid in a circular tube, stating all assumptions.
2. Compare power-law, Carreau, Herschel–Bulkley, and Oldroyd-B models in terms of physical meaning and applicability.
3. Explain how low Reynolds number, diffusion, surface tension, and electrokinetics influence microfluidic design.
4. Discuss the role of the Womersley number and vessel compliance in pulsatile cardiovascular flow.
5. Describe a rigorous workflow for selecting, calibrating, verifying, and validating a non-Newtonian CFD model.

Analytical / Application Questions

1. A process engineer doubles the radius of a power-law-fluid tube while keeping pressure gradient, length, K, and n fixed. Determine how the ideal flow rate scales with radius and explain the practical significance.
2. A microchannel has large Péclet number but a short mixing length. Should rapid axial transport be interpreted as rapid mixing? Analyze the roles of transverse diffusion and channel geometry.
3. A blood-analysis cartridge produces different pressure drops for samples with equal flow rate but different hematocrit. Propose a dimensionless and experimental investigation that separates rheological, geometric, and fouling effects.

Multiple Choice Questions with Answers

Question 1

For a power-law fluid, what does n < 1 indicate?

A. Newtonian behavior

B. Shear-thinning behavior

C. Shear-thickening behavior

D. Ideal elastic-solid behavior

Correct Answer: B. Shear-thinning behavior

Question 2

Which quantity is independent of the constitutive law for fully developed pressure-driven flow in a circular tube?

A. Velocity profile

B. Volumetric flow rate

C. Radial shear-stress distribution

D. Apparent viscosity

Correct Answer: C. Radial shear-stress distribution

Question 3

The capillary number compares which two effects?

A. Inertia and gravity

B. Viscous stress and surface tension

C. Diffusion and reaction

D. Elasticity and compressibility

Correct Answer: B. Viscous stress and surface tension

Question 4

If the Péclet number is large, which statement is generally correct?

A. Molecular diffusion dominates advection

B. Advection dominates molecular diffusion

C. Surface tension is necessarily negligible

D. The flow must be turbulent

Correct Answer: B. Advection dominates molecular diffusion

Question 5

For a Newtonian fluid in a circular tube, doubling the tube radius while holding pressure gradient, viscosity, and length fixed changes Q by what factor?

A. 2

B. 4

C. 8

D. 16

Correct Answer: D. 16

Question 6

What does a large Weissenberg number generally suggest?

A. Elastic effects may be important

B. Molecular diffusion dominates

C. The fluid has zero viscosity

D. The flow is necessarily turbulent

Correct Answer: A. Elastic effects may be important

Question 7

Which model includes a yield stress and a power-law post-yield response?

A. Carreau model

B. Herschel–Bulkley model

C. Newtonian model

D. Ideal-gas model

Correct Answer: B. Herschel–Bulkley model

Question 8

In ideal creeping flow, the Reynolds number is typically

A. Much greater than one

B. Approximately zero or much less than one

C. Exactly equal to one

D. Undefined because density is zero

Correct Answer: B. Approximately zero or much less than one

Question 9

Which phenomenon is associated with small-vessel blood flow?

A. Fåhræus–Lindqvist effect

B. Venturi effect only

C. Doppler-freezing effect

D. Rayleigh–Taylor equilibrium

Correct Answer: A. Fåhræus–Lindqvist effect

Question 10

In the Helmholtz–Smoluchowski expression, electro-osmotic velocity depends directly on which quantity?

A. Electric-field strength

B. Pipe roughness only

C. Gravitational acceleration only

D. Turbulent kinetic energy

Correct Answer: A. Electric-field strength