Fingering Instability in Hele-Shaw cell · Bischofberger, I.; Ramachandran, R. & Nagel, S. R.,...
Transcript of Fingering Instability in Hele-Shaw cell · Bischofberger, I.; Ramachandran, R. & Nagel, S. R.,...
Fingering Instability in Hele-Shaw cell
Lucas Amoudruz
Computational Science and Engineering Laboratory
with:
Sergey LitvinovPetros Koumoutsakos
OUTLINE
• Fingering instabilities in Hele-Shaw cells • Linear stability analysis in 2D • Simulations with Dissipative Particle Dynamics • Results
Fingering instabilities
- low viscosity fluid flowing inside high viscosity - same densities
Credits: Nagel Group, University of Chicago
- oil recovery - sugar refining - fundamental question: instabilities
Hele Shaw flow
b
L
∇p
Re ≪ 1 ∇ ⋅ u = 0Stokes approx: μΔu = ∇p
Darcy + continuity 2D potential flow: Δϕ = 0
μviscosity
bL
≪ 1 Darcy’s law: u =κμ
∇p = ∇ϕ
porosity κ =b2
12
Stability analysis, 2D
perturbate interface: x = α exp (iky + σt)
Saffman, P. G. & Taylor, G. The penetration of a fluid into a porous medium or Hele-Shaw cell containing a more viscous liquid, 1958
no surface tension: p− = p+ (at interface)
σkκ (μ+ + μ−) = (μ+ − μ−) U always unstable for μ− < μ+
x
y
μ− μ+
U
suggests 3D effectscontradicts experiments
Available numerical models?
Tryggvason, G. & Aref, H. Numerical experiments on Hele Shaw flow with a sharp interface
• Continuum methods: Boundary Integral, Level sets • 2D simulations • Previous work did not consider zero surface tension limit
Li, S.; Lowengrub, J. S. & Leo, P. H. A rescaling scheme with application to the long-time simulation of viscous fingering in a Hele--Shaw cell
Dissipative Particle Dynamics (DPD)
• linear and angular momentum conserved • intrinsic fluctuations, suitable for stability analysis • efficient implementation: uDeviceX (GB15 finalist)
Dissipative Particle Dynamics
Fi = ∑j≠i
(FCij + FD
ij + FRij ) eij
∂ri
∂t= vi
∂vi
∂t=
1m
Fi
FCij = awC(rij)
FDij = − γwD(rij)(rij ⋅ vij)
FRij = σwR (rij) ζij
σ2 = 2kBTγ
rc
rij ij
Solvent Wall
fluctuation-dissipation theorem:
Groot, R. D.; Warren, P. B. & others: Dissipative particle dynamics: Bridging the gap between atomistic and mesoscopic simulation, 1997
⟨ξij(t)ξkl(t′�)⟩ = δ(t − t′�)(δikδjl + δilδjk)⟨ξij(t)⟩ = 0
repulsion: pressure
dissipation: viscosity
fluctuations
wD = (wR)2
Boundary conditions
Groot, R. D.; Warren, P. B. & others: Dissipative particle dynamics: Bridging the gap between atomistic and mesoscopic simulation, 1997
• Interactions with frozen particles • Bounce back on wall surface
SolventWall
(r0, v0)
(r1, v1)
(rn, vn)
(rw, vw)
12 (v0 + vn) = vw
vn = 2vW − v0rn = rw + (dt − τ)vn
Collision time τ ∈ [0, dt]
new velocity such that
DPD: validation
4 roller mill Poiseuille Couette Startup
∇p
U0
∇p
High performance software: uDeviceX
• High performance and High throughput solver for N-body, short range interactions • MPI + CUDA implementation
• 2’097'152 particles per node • ~34 neighbours per particle • 7.6 ms per time step
Rossinelli, et al., 2015 The in-silico lab-on-a-chip: petascale and high-throughput simulations of microfluidics at cell resolution (GB 15 Finalist)
Computational Setup
∇p
Solvent 1
Solvent 2
• periodic boundary conditions • label particles crossing outflow boundary 2 → 1
DPD: multiple solvents
a11 = a12 = a22
γ12 =γ11 + γ22
2
no interfacial tension:
solvent 1
solvent 2
two solvents two kinds of particles
dissipative force parameter
Results: fingering patterns
μ−
μ+= 0.5
μ−
μ+= 0.0625
μ−
μ+= 0.125
μ−
μ+= 0.05
Always stable for μ−
μ+≥ 0.33
μ−
μ+≥ 1Recall: stability analysis in 2D: stable only for
Simulation results
RfRi
μ−μ+
Bischofberger, I.; Ramachandran, R. & Nagel, S. R., Fingering versus stability in the limit of zero interfacial tension, 2014
SUMMARY
• DPD able to reproduce fingering instabilities • First simulation to capture phase transition as in experiments • Difficulties for very small viscosity ratios: density fluctuations too large • First step to understand this phenomenon