Deterministic Micro-Flow: Variational & Kinetic Models

• Deterministic Micro-Flow (DMF) is a framework of noise-free methods that use deterministic particle and kinetic schemes to model microflow systems without stochastic errors.
• DMF employs structure-preserving variational discretizations and energy decay principles to accurately simulate viscoelastic, polymeric, and rarefied flows.
• Applications include micro-pump drift analysis and unified gas-kinetic schemes, offering computationally efficient, multiscale solutions for complex microfluidic systems.

Deterministic Micro-Flow (DMF) encompasses a spectrum of mathematically rigorous, noise-free methodologies for modeling, simulation, and analysis of microflow systems wherein the underlying particle, kinetic, or hydrodynamic processes are encoded in fully deterministic (non-Monte Carlo) algorithms. These approaches span deterministic particle-based schemes for micro-macro viscoelastic flows, structure-preserving variational discretizations, and velocity-space-resolved solvers for rarefied and continuum regimes. DMF frameworks support accurate analysis of rheological effects, transport characteristics, and flow–structure interactions relevant in microfluidics, micro-pumps, and polymeric fluid dynamics (Bao et al., 2021, Beltrame et al., 2012, Liu et al., 2020).

1. Deterministic Micro-Flow: Context and Core Principles

Deterministic Micro-Flow is defined by schemes that discretize the equations governing microflows (including both continuum and kinetic descriptions) using algorithms that are strictly deterministic, eschewing the stochasticity of Monte Carlo or Brownian sampling. In polymeric and viscoelastic flows, this typically means replacing stochastic trajectory ensembles by structure-preserving, variational particle systems; in kinetic theory, deterministic quadratures in velocity space supplant random or particle-based methods. DMF models are underpinned by:

• Conservation and dissipation principles (e.g., energy-decay laws in viscoelastic models)
• Thermodynamically-consistent variational formulations (e.g., discrete energetic variational approaches)
• Direct deterministic discretization of distribution functions and phase-space (velocity, configuration) (Bao et al., 2021, Liu et al., 2020)

Applications span drift phenomena in micro-pumps, transitional/rarefied regime gas flows, and micro–macro coupled viscoelastic simulations.

2. Governing Equations and Deterministic Reduction

In classical micro–macro models (polymeric liquids), the coupled system consists of a macroscopic incompressible Navier–Stokes equation and a kinetic equation for the microscopic configuration density $f(x,q,t)$. The governing equations include:

• Incompressibility: $\nabla_x\cdot u = 0$
• Navier–Stokes with polymer stress: $$\rho(u_t + u\cdot\nabla_x u) + \nabla_x p = \eta_s \Delta_x u + \nabla_x\cdot\tau$$
• Kinetic (Fokker–Planck) equation:

$$f_t + \nabla_x\cdot(u f) + \nabla_q\cdot\bigl((\nabla_xu)\,q\,f\bigr) = \frac{2}{\zeta}\nabla_q\cdot(f\,\nabla_q\Psi) + \frac{2k_BT}{\zeta}\Delta_q f$$

• In micro-pump drift models, particle motion is governed by linear Stokes drag in the overdamped limit:

$\dot{x}_p = u_0(x_p) \sin (2\pi t)$

reflecting a 1D deterministic ratchet for particle transport in oscillatory flows (Beltrame et al., 2012).

Deterministic reduction in DMF replaces the continuous or stochastic kinetic distributions with particle- or velocity-grid-based deterministic representations: e.g., weighted Dirac masses in configuration space, or deterministic quadrature nodes in velocity space for kinetic equations (Bao et al., 2021, Liu et al., 2020).

3. Structure-Preserving Deterministic Particle Schemes

DMF realizes the particle-based discretization of continuous distribution functions—for example, in viscoelastic micro-macro models:

$$f(x,q,t) \approx \frac{1}{N}\sum_{i=1}^N\delta(q - q_i(x,t))$$

where $\{q_i\}$ are deterministic particle states, each associated with a mesh node or spatial location. The evolution equations for the particles are obtained via a discrete energetic variational approach (EnVarA):

• Least Action Principle applied to the discrete kinetic/free energy yields conservative forces
• Maximum Dissipation Principle generates dissipative forces
• The resulting ODE system preserves discrete energy-decay:

$$\dot{q}_i(x,t) = (\nabla_x u)q_i - \frac{2}{\zeta} \left[ k_BT [\ldots] + \nabla_{q_i}\Psi(q_i) \right]$$

with additional kernel regularization for entropy and free energies to manage the singularity of empirical measures (Bao et al., 2021).

Time discretization is performed using incremental variational minimization at each time step, ensuring unconditional energy decay at the discrete level. Convection in space is handled via operator splitting and Lagrangian transport of the computational mesh.

4. Deterministic Kinetic Solvers: Unified Gas-Kinetic Scheme

The Unified Gas-Kinetic Scheme (UGKS) provides a deterministic framework for micro-flow simulations across the entire Knudsen number regime. The approach starts from the (linearized) BGK kinetic model:

$$\frac{\partial f}{\partial t} + \mathbf{v}\cdot\nabla_{\mathbf{x}}f = \frac{g-f}{\tau}$$

where $f$ is the velocity distribution, $g$ the local Maxwellian, and $\tau$ the relaxation time. DMF here is realized via:

• Deterministic velocity discretization: uniform Cartesian grids in high-Kn regimes or Gauss-Hermite quadrature for fluid-like regimes
• Analytic integral solutions for cell interface fluxes, capturing true multiscale fluxes across all regimes
• Simultaneous evolution of microscopic (distribution $f$) and macroscopic (conservative variables) quantities, closed via moment relations
• Asymptotic preservation: recovers linearized Navier–Stokes flux in the continuum limit and free-molecular flux in the kinetic limit (Liu et al., 2020)

This approach satisfies the "Unified Preserving" (UP) property, meaning that the scheme rigorously preserves Chapman–Enskog expansions up to a given order for time and grid steps much larger than kinetic mean free paths, enabling accurate resolution of viscous boundary layers and other multiscale features.

5. Deterministic Transport and Bifurcation Mechanisms in Micro-Pumps

Deterministic modeling of micro-pumps reveals two distinct mechanisms for net particle transport in periodic, oscillatory, viscous flow fields:

• Spatio-temporal synchronization (velocity locking):
  • Phase-locked transport solutions exist for integer drift speeds $c\in\mathbb{Z}$, corresponding to exactly one pore position advanced per oscillation ($v_\text{drift}=L/T$).
  • Emerges via the interaction of periodic orbits in the comoving phase space; robust in narrow parameter bands in $(u_m,\gamma)$, sensitive to driving frequency and flow amplitude.
• Intermittent (phase-slip) drift:
  • Occurs when synchronized periodic solutions become unstable via saddle-node bifurcation.
  • The particle trajectory consists of long oscillatory laminar phases, punctuated by rapid pore-to-pore jumps—a type-I intermittent bifurcation (Manneville–Pomeau).
  • The mean drift near onset follows a square-root scaling law $$|v_\text{drift}| \propto (\text{parameter} - \text{critical})^{1/2}$$.
  • Dominant in large-drag and moderate-asymmetry conditions, with transport direction set by pore shape (Beltrame et al., 2012).

These mechanisms are deterministic, arising from the fine structure of the underlying nonlinear equations, and can be mapped out by bifurcation diagrams and stability analysis.

6. Numerical Implementation, Algorithmic Flow, and Validation

DMF numerical algorithms consist of:

1. Macro-step: finite-element discretization of macroscopic (Navier–Stokes-type) variables (e.g., isoP2/P1, or pressure-correction schemes).
2. Micro-step: solution of deterministic particle ODE systems via variational minimization (for micro-macro models) or deterministic quadrature (for kinetic models).
3. Stress or flux interpolation: evaluation of macroscopic feedback terms (e.g., polymeric stress from particles, or fluxes from distribution functions) and projection onto discrete spaces.
4. Operator splitting or Lagrangian mesh update for convective terms.
5. Parameter regimes typically involve tens to hundreds of mesh points and $N=50$–$500$ particles or quadrature points; kernel bandwidths are chosen based on problem geometry and smoothing requirements (Bao et al., 2021, Liu et al., 2020).

Numerical results for DMF include:

• Accurate reproduction of analytical solutions for benchmark flows (e.g., Oldroyd-B Couette, FENE startup, micro-Poiseuille)
• Resolution of rheological features such as hysteresis, $\delta$-spike configuration PDFs, shear overshoot, symmetry breaking, and vortex migration
• Capability of resolving high-Kn (rarefied) and continuum regimes without stochastic noise, using moderate numbers of particles or velocity points
• Efficient capture of viscous boundary layers even for coarse spatial grids, as dictated by the UP property

7. Advantages, Limitations, and Prospects

Deterministic micro-flow methodologies confer several advantages:

• Strictly noise-free simulations, obviating the need for variance reduction techniques
• Fidelity to the energetic and variational structure of the continuous models, guaranteeing discrete energy decay and conservation laws
• Low numerical overhead: modest particle/quadrature counts suffice for benchmark accuracy relative to stochastic approaches
• Naturally extendable to nontrivial rheological models (e.g., FENE, Doi–Onsager, multichain systems) and kinetic systems in complex microfluidic domains

Limitations include the need for careful regularization parameters (kernel bandwidth) in particle-based entropy approximations, potential splitting errors from operator splitting, and computational cost increases for high-dimensional configuration spaces. Prospective developments include spatially adaptive kernels, advanced stress reconstruction from particle ensembles, direct coupling without operator splitting, extensions to multi-component and reactive systems, and fully parallelized, mesh-adaptive implementations (Bao et al., 2021).

Deterministic Micro-Flow provides a rigorous, variance-free, and structure-preserving toolkit for analysis, modeling, and simulation of microflows in both viscoelastic and kinetic regimes, with demonstrated applicability to systems requiring accurate micro–macro coupling, bifurcation analysis, and multiscale computational efficiency (Bao et al., 2021, Liu et al., 2020, Beltrame et al., 2012).