| name | fea-fluid-structure |
| description | Fluid-structure interaction (FSI) FEA — ALE formulation, added mass, one-way vs. two-way coupling, vortex-induced vibration, hydroelastic flutter, aeroelastic divergence, ABAQUS/ANSYS/OpenFOAM coupling, pressure mapping. |
| metadata | {"priority":7,"promptSignals":{"phrases":["fluid structure interaction","FSI analysis","aeroelastic","hydroelastic","ALE formulation","added mass FEA"],"minScore":3}} |
Fluid-Structure Interaction (FSI) FEA — Complete Skill
FSI Classification
One-Way (Uncoupled) FSI
Method: CFD → extract pressure field → apply as load to FEA (no feedback)
Applicable when: structural deformation << fluid domain scale; deformation does not significantly change fluid flow
Examples: static pressure loading on aircraft panel; wind load on building facade; pump impeller under steady pressure
Workflow: CFD (OpenFOAM/ANSYS Fluent) → pressure file → FEA (ABAQUS/ANSYS) surface pressure load
Interpolation: nearest-neighbor or RBF (radial basis function) mapping of pressure from CFD mesh to FEA mesh
Two-Way (Coupled) FSI
Method: fluid and structure solvers exchange pressure, displacement, velocity iteratively each time step
Applicable when: structural deformation significantly affects flow (flutter, large-deflection membranes, biological valves)
Examples: heart valve; bridge flutter; compliant pipe; flexible wing at high angle of attack
Coupling methods:
- Partitioned: separate fluid (CFD) + structure (FEA) solvers with data exchange at interface
- Monolithic: single solver matrix for fluid+structure DOF (more stable but expensive)
Arbitrary Lagrangian-Eulerian (ALE) Formulation
Problem with pure Lagrangian (FEA mesh follows material): mesh distorts severely with large fluid deformation
Problem with pure Eulerian (CFD fixed mesh): material interface tracking difficult
ALE solution: mesh moves arbitrarily (neither with material nor fixed); decouple mesh motion from material motion
ALE governing equation (convective form):
ρ(∂v/∂t|_χ + (v - v̂)·∇v) = ∇·σ + ρb
χ = ALE reference coordinate; v̂ = mesh velocity; (v - v̂) = convective velocity
Mesh motion strategies:
- Laplacian smoothing: solve ∇²x_mesh = 0 in fluid domain; maintains mesh quality
- Spring analogy: mesh elements as springs; stiffer near boundaries
- Remeshing: when quality degrades below threshold → interpolate fields to new mesh
ABAQUS ALE adaptive meshing: *ADAPTIVE MESH CONTROLS; use for impact and large-deformation problems
Added Mass Effect
Phenomenon: fluid resists acceleration of submerged structure → effective mass increases
Added mass coefficient C_a (potential flow):
- Circular cylinder (transverse): m_a = C_a × ρ_f × π/4 × D² × L; C_a = 1.0 (theory)
- Sphere: m_a = C_a × ρ_f × (2/3)π r³; C_a = 0.5
- Flat plate (normal): C_a ≈ (π/4) × D²/L (aspect ratio dependent)
Effect on natural frequency:
ω_wet = ω_dry / √(1 + m_a/m_structure)
For submerged pipe: f_wet / f_dry ≈ 0.5–0.8 depending on fluid/structure density ratio
Implementation: add mass matrix M_added = C_a × ρ_f × V_displaced to structural mass matrix
Vortex-Induced Vibration (VIV)
Strouhal number: St = f_s × D / U_∞ [f_s = vortex shedding frequency; D = diameter; U = free stream velocity]
St ≈ 0.2 for circular cylinder (Re = 1000–10⁵)
Lock-in condition: when f_s ≈ f_n (structural natural frequency) → resonance → large amplitude vibration
Lock-in range: 0.7 ≤ U_r ≤ 1.4 [U_r = U/(f_n × D) = reduced velocity]
Amplitude at lock-in:
A/D ≈ 1.0 for lightly damped cylinders in air; A/D ≈ 0.5–0.8 for water
Scrutan number: Sc = 2m_eff × ζ / (ρ_f × D²) [controls VIV amplitude; Sc > 50 → no significant VIV]
FEA-CFD coupling for VIV:
- Start CFD with stationary cylinder → extract forces
- Apply forces to structural FEA → compute displacement
- Update CFD mesh to match structural displacement
- Repeat each time step (typical dt = 0.01 × period)
Hydroelastic Flutter
Critical flutter velocity (panel in flow):
U_flutter = U_divergence / √2 [approximate for 2D panel]
Flutter condition: coupled bending-torsion mode → imaginary part of frequency eigenvalue becomes zero
Theodorsen function approach (2D airfoil flutter):
Matrix: [M]{ẍ} + [C]{ẋ} + [K]{x} = q_dyn × [A(k)]{x} [k = reduced frequency = ωb/U]
p-k method or v-g method to find flutter speed and flutter frequency
Hydroelastic divergence:
Divergence: static instability; elastic restoring moment < hydrodynamic moment
q_divergence = 1/2 × ρ × U² → K_T / (∂C_L/∂α) × (∂α/∂δ)⁻¹
ABAQUS/ANSYS FSI Coupling
ABAQUS + ABAQUS/CFD (co-simulation):
*CO-SIMULATION: link ABAQUS structural + CFD solver
Interface: *CO-SIMULATION REGION; fluid-structure boundary
Exchange: displacement (structure → fluid); pressure/traction (fluid → structure)
Staggered scheme: structure predictor → fluid correction → update
ANSYS Mechanical + ANSYS Fluent (System Coupling):
Workbench: drag Structural and Fluid components into System Coupling cell
Interface: Named Selection at FSI surface; auto-mapping of pressure and displacement
Explicit (1-way) or implicit (2-way) coupling; subcycling allowed
OpenFOAM + structural solver (FEniCS/Code_Aster):
preCICE coupling library: surface mesh exchange; RBF interpolation; convergence monitoring
solidDisplacementFoam + FSI adapter for OpenFOAM solid solver
Pressure Mapping (One-Way)
CFD to FEA pressure transfer:
- Extract pressure on FSI surface from CFD (surface mesh)
- Interpolate to FEA mesh (different node locations)
- Apply as distributed pressure load in FEA
Interpolation methods:
- Nearest node: fast; inaccurate for coarse-to-fine mapping
- RBF (radial basis function): accurate; handles non-matching meshes; computationally expensive
- Consistent pressure interpolation: preserve total force: F_FEA = ∑ p_FEA × A_FEA ≈ F_CFD
ABAQUS pressure loading:
*DSLOAD, OP=NEW: apply distributed surface load
Normal mapping: MAP SOLUTION between CFD and FEA (ABAQUS mapping tool)
Convergence and Stability
Partitioned coupling stability:
Strongly coupled: use Aitken relaxation to accelerate convergence: ω = ω_old × r_old/(r_old - r_new) [r = residual]
Weak coupling: explicit (one exchange per time step) → unstable for ρ_fluid ≈ ρ_structure (added mass instability)
Time step for explicit FSI:
Δt_FSI ≤ min(Δt_fluid_CFL, Δt_structure_CFL, Δt_interface)
For implicit: Δt can be larger; stability via Newton iteration within each time step
Convergence criterion:
Displacement residual: ||x_new - x_old|| / ||x_new|| < 10⁻⁴
Force residual: ||F_fluid - F_structure|| / ||F_fluid|| < 10⁻³
Validation Benchmarks
Turek-Hron FSI benchmark: flexible beam behind square cylinder; well-documented; compare tip displacement and frequency
NACA airfoil flutter: compare computed flutter speed to linear theory
Circular cylinder VIV: compare A/D vs. reduced velocity to Gopalkrishnan (1993) data
Standards
| Standard | Scope |
|---|
| ASME V&V 10 | Verification and validation in computational solid mechanics |
| ASME V&V 20 | V&V for CFD |
| ISO 7626 | Mechanical vibration — structural dynamics measurements |
| ASME PTC 19.3 | Temperature measurement in flow |
Output
Provide: FSI type (one-way/two-way), coupling method (partitioned/monolithic), mesh motion scheme (ALE/remeshing), structural solver (ABAQUS/ANSYS), fluid solver (ANSYS Fluent/OpenFOAM), time step Δt [s] and number of coupling iterations per step, maximum structural displacement [mm], critical flutter velocity U_flutter [m/s] (if applicable), VIV lock-in check (reduced velocity U_r and Scrutan number Sc), added mass ratio m_a/m_structure, peak pressure on FSI surface [MPa], pressure mapping error (force balance check [%]), convergence achieved (displacement residual), and applicable standard (ASME V&V 10, V&V 20).