UNIT 3: SOLUTION METHODOLOGIES, ANALYSIS TYPES & POST-PROCESSING
3.1 Core Finite Element Solution Procedures
3.1.1 Static Structural Analysis
-
Direct Stiffness Method (Assembly, BCs, Solution):
-
Element Stiffness Matrix: For each element, derive $$\displaystyle [k^e] $$ from material properties ($E$, $\nu$) and geometry.
-
Assembly: Assemble global stiffness matrix $[K]$ and global force vector ${F}$ from all $$\displaystyle [k^e] $$ and $$\displaystyle {f^e} $$.
-
Apply Boundary Conditions: Modify $[K]$ and ${F}$ to enforce constraints (e.g., fixed displacements).
-
Solve: Solve the system $$\displaystyle [K]\{u\} = \{F\} $$ for nodal displacements $\{u\}$.
-
Post-process: Compute strains $\{\varepsilon\}$ and stresses $\{\sigma\}$ from $\{u\}$.
Core Equation: $$\displaystyle \boxed{[K]\{u\} = \{F\}} $$
-
-
Linear vs. Nonlinear Static Analysis:
-
Linear: $[K]$ is constant. Superposition applies. Fast convergence.
-
Nonlinear: $[K]$ changes with deformation. Requires iterative solvers (Newton-Raphson).
-
Geometric Nonlinearity: Large deformations/displacements (e.g., buckling, snap-through).
-
Material Nonlinearity: Stress-strain curve is nonlinear (e.g., plasticity, hyperelasticity).
-
Contact Nonlinearity: Surfaces can separate, slide, or have friction.
-
-
-
Contact Analysis:
-
Types: Surface-to-Surface (more accurate, robust), Node-to-Surface (simpler, less accurate).
-
Friction Models: Coulomb friction (static/kinetic coefficients), frictionless.
-
Key Parameters: Contact stiffness, penetration tolerance, stabilization.
[!TIP] In lab exams, you may be asked to set up a simple contact pair (e.g., indentation) and interpret contact pressure plots.
-
3.1.2 Dynamic Analysis
-
Modal Analysis (Eigenvalue Problem):
-
Solves for natural frequencies $$\displaystyle \omega_n $$ and mode shapes $$\displaystyle \phi_n $$.
-
Equation: $$\displaystyle ([K] - \omega_n^2[M])\{\phi_n\} = \{0\} $$
-
Purpose: Avoid resonance, understand vibration characteristics.
-
Output: Frequency table, animated mode shapes.
-
-
Harmonic Response Analysis:
-
Forced vibration at specific, steady-state frequencies.
-
Solves $$\displaystyle (-\omega^2[M] + i\omega[C] + [K])\{U\} = \{F_0\} $$ for complex displacement amplitude $\{U\}$.
-
Output: Displacement/Stress vs. Frequency (bode plot).
-
-
Transient Dynamic Analysis:
-
Time-dependent response to arbitrary loads.
-
Explicit (e.g., LS-DYNA): Conditionally stable (CFL limit), no matrix inversion, good for short-duration, high-speed events.
-
Implicit (e.g., ANSYS Mechanical): Unconditionally stable, requires matrix inversion (Newton iterations), good for long-duration, quasi-static events.
Key Difference: Modal finds free vibration characteristics. Harmonic finds steady-state forced response at chosen frequencies. Transient finds full time-history response.
-
3.1.3 Thermal Analysis
-
Steady-State Heat Transfer:
-
$$\displaystyle \nabla \cdot (k\nabla T) + Q = 0 $$ (no time term).
-
Boundary Conditions (BCs):
-
Temperature: Specified $$\displaystyle T = T_0 $$.
-
Convection: $$\displaystyle -k\frac{\partial T}{\partial n} = h(T - T_{\infty}) $$.
-
Radiation: $$\displaystyle -k\frac{\partial T}{\partial n} = \epsilon \sigma (T^4 - T_{\infty}^4) $$ (often linearized).
-
-
Output: Temperature field $T(x,y,z)$, heat flux.
-
-
Transient Heat Transfer:
-
$$\displaystyle \rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k\nabla T) + Q $$.
-
Requires initial temperature field $$\displaystyle T(t=0) $$.
-
Output: Temperature vs. time at points, thermal wave propagation animation.
-
3.2 Computational Fluid Dynamics (CFD) Solution Fundamentals
3.2.1 Governing Equations & Discretization
-
Navier-Stokes Equations (Conservation Laws):
-
Mass (Continuity): $$\displaystyle \frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \vec{V}) = 0 $$
-
Momentum: $$\displaystyle \rho \left( \frac{\partial \vec{V}}{\partial t} + \vec{V} \cdot \nabla \vec{V} \right) = -\nabla p + \nabla \cdot \left( \mu \nabla \vec{V} \right) + \vec{f} $$
-
Energy: $$\displaystyle \rho c_p \left( \frac{\partial T}{\partial t} + \vec{V} \cdot \nabla T \right) = \nabla \cdot (k \nabla T) + \Phi $$
-
-
Finite Volume Method (FVM) Overview:
-
Integrates governing equations over control volumes (cells).
-
Transforms volume integrals to surface integrals via Gauss theorem.
-
Core Idea: $$\displaystyle \int_V \frac{\partial (\rho \phi)}{\partial t} dV + \int_S \rho \phi \vec{V} \cdot d\vec{S} = \int_V \nabla \cdot (\Gamma \nabla \phi) dV + \int_V S_\phi dV $$
-
Fluxes through cell faces are approximated (e.g., central differencing, upwind).
-
-
Pressure-Velocity Coupling:
-
Problem: Pressure appears in momentum equations but has no own equation.
-
Algorithms:
-
SIMPLE (Semi-Implicit Method for Pressure-Linked Equations): Guess pressure → solve momentum → correct pressure/velocity. Standard for steady-state.
-
SIMPLER (SIMPLE Revised): Separate pressure correction step, often more robust.
-
PISO (Pressure-Implicit with Splitting of Operators): For transient, does multiple corrector steps per timestep.
-
| Algorithm | Primary Use | Key Feature | | :--- | :--- | :--- | | SIMPLE | Steady-State | One pressure correction per iteration | | PISO | Transient | Multiple corrector steps per timestep |
-
3.2.2 Turbulence Modeling (RANS Approach)
-
Need: Direct Numerical Simulation (DNS) of turbulence is computationally prohibitive for most engineering flows.
-
RANS (Reynolds-Averaged Navier-Stokes): Decompose $$\displaystyle u_i = \overline{u_i} + u_i' $$. Solve for time-averaged $$\displaystyle \overline{u_i} $$, model Reynolds stresses $$\displaystyle -\rho \overline{u_i' u_j'} $$.
| Model | Equations | Best For | Limitations | | :--- | :--- | :--- | :--- | | k-ε | 2 (k, ε) | External flows, general purpose | Poor near walls, poor for rotation/curvature | | k-ω SST | 2 (k, ω) | Adverse pressure gradients, separation | Can be over-sensitive to freestream ω | | Spalart-Allmaras | 1 (ν̃) | Aerodynamic flows (wall-bounded) | Less general, tuned for aerospace |
3.2.3 Boundary Conditions for CFD
-
Inlet:
-
Velocity Inlet: Specify $\vec{V}$ (and often turbulence parameters).
-
Pressure Inlet: Specify total pressure (for compressible) or gauge pressure (for incompressible).
-
Mass Flow Inlet: Specify mass flow rate.
-
-
Outlet: Pressure Outlet (most common). Specify static pressure; flow direction extrapolated.
-
Wall: No-slip ($$\displaystyle \vec{V}=0 $$) for viscous flows. Can specify wall motion, roughness, thermal condition (temperature, heat flux).
-
Symmetry: Mirror plane; zero normal velocity, zero normal gradients for other variables.
-
Periodic: For repeating geometries (e.g., tube bundles).
-
Pressure Far-Field: For external aerodynamics (specifies freestream conditions at infinity).
3.3 Convergence, Accuracy & Numerical Stability
3.3.1 Convergence Criteria
-
Residual Monitoring:
-
Residual = Imbalance in conservation equations (mass, momentum, energy).
-
Absolute: Raw value. Hard to judge.
-
Relative: $$\displaystyle \frac{R_{current}}{R_{initial}} $$. Common target: $$\displaystyle < 10^{-3} $$ to $$\displaystyle 10^{-5} $$.
-
Normalized: Scaled by cell volume/flux. Most software default.
-
-
Solution Convergence vs. Grid Independence:
-
Convergence: Iterations lead to a stable solution (residuals plateau, monitors flat).
-
Grid Independence Study: CRITICAL FOR VALIDATION. Refine mesh until key results (e.g., $$\displaystyle C_d $$, max temperature) change by less than an acceptable threshold (e.g., 1-2%).
[!TIP] Never trust results from a single mesh. Always perform a grid independence study for reports and viva.
-
3.3.2 Error Sources & Control
-
Discretization Error: Due to approximating derivatives (finite differences/volumes). Reduced by mesh refinement.
-
Iteration Error: Incomplete convergence of nonlinear/implicit solvers. Reduced by tighter convergence criteria.
-
Round-off Error: Finite precision arithmetic. Usually negligible.
-
Under-Relaxation Factors (URFs):
-
Purpose: Dampen variable updates to stabilize convergence.
-
Tuning: Start with default (e.g., 0.3-0.7 for pressure, 0.5-0.8 for momentum). Increase gradually for faster convergence; decrease if diverging.
-
3.3.3 Stability Analysis
-
CFL (Courant-Friedrichs-Lewy) Condition:
-
For explicit schemes: $$\displaystyle CFL = \frac{u \Delta t}{\Delta x} \leq C_{max} $$.
-
$$\displaystyle C_{max} $$ depends on scheme (often $\leq 1$). Violation leads to instability.
-
Implication: $\Delta t$ must be small enough or $\Delta x$ fine enough.
-
-
Mesh Quality: High skewness, non-orthogonality, high aspect ratio can degrade accuracy and stability. Check mesh quality metrics.
3.4 Post-Processing & Results Interpretation
3.4.1 Contour/Vector/Deformation Plots
-
Contour Plots: Show scalar field distribution (Stress $\sigma$, Temperature $T$, Pressure $p$, Velocity magnitude $|\vec{V}|$). Use appropriate smoothing/contour levels.
-
Vector Plots: Show direction and magnitude of vector fields (Velocity $\vec{V}$, displacement $\vec{u}$). Use arrow scaling.
-
Deformation Plots:
-
True Scale: Deformation is actual magnitude. Often too small to see.
-
Exaggerated Scale: Multiply displacements by a scale factor (e.g., 10x, 100x) for visibility. Always state the scale factor used.
-
3.4.2 Quantitative Data Extraction
-
Probes/Point History: Monitor a variable (e.g., temperature, pressure) at a specific node/cell over time/iterations. Plot as XY graph.
-
Path Lines/Streamlines:
-
Path Lines: Trajectory of fluid particles over time (transient).
-
Streamlines: Instantaneous flow direction (steady-state).
-
-
Report Generation:
-
Reaction Forces/Moments: At constrained boundaries.
-
Maximum/Minimum Values: e.g., Max von-Mises stress, Max temperature.
-
Averages/Integrals: e.g., Average pressure on a surface, total heat flux.
-
Fluxes: Mass flow rate at inlets/outlets (should balance).
-
3.4.3 Animation
-
Deformation Animation: For static nonlinear or transient structural results.
-
Flow Animation: For transient CFD (velocity vectors, contours evolving).
-
Purpose: Visualize physical phenomena (vibration mode, vortex shedding, thermal wave).
3.5 Verification & Validation (V&V) in Simulation
| Aspect | Verification (V) | Validation (V) |
|---|---|---|
| Question | "Are we solving the equations right?" | "Are we solving the right equations?" |
| Goal | Check for coding/implementation errors, numerical accuracy. | Check if model predictions match physical reality. |
| Activities | 1. Code Verification: Test solver against known analytical solutions (method of manufactured solutions).<br>2. Calculation Verification: Estimate discretization error via grid convergence study (Richardson extrapolation). | 1. Compare simulation results with experimental data or high-fidelity benchmarks.<br>2. Assess physics fidelity (e.g., correct turbulence model, material model, BCs). |
| Output | Error estimates, order of accuracy, confidence in numerical solution. | Quantitative measure of model accuracy (e.g., error %), qualitative assessment of model adequacy. |
Key Point: Verification is about numerical correctness. Validation is about physical correctness. Both are mandatory for credible simulation.
3.6 Advanced & Specialized Topics
3.6.1 Multiphysics Coupling
-
Fluid-Structure Interaction (FSI):
-
One-Way: Fluid flow affects structure (pressure loading), but structure deformation does not affect flow. Simpler, sequential.
-
Two-Way: Fully coupled. Fluid pressure deforms structure, which in turn changes fluid domain (mesh motion). Requires iterative solver coupling (e.g., ANSYS System Coupling).
-
-
Thermo-Mechanical Analysis:
- Thermal Stress: Temperature field from thermal analysis is mapped as a load (thermal strain $$\displaystyle \varepsilon_{th} = \alpha \Delta T $$) in a subsequent structural analysis.
-
Other Examples: Piezoelectric (structural stress → electric potential), Joule Heating (current → heat).
3.6.2 Optimization & Parametric Studies
-
Design of Experiments (DOE): Systematic way to vary input parameters (e.g., thickness, inlet velocity) to explore design space.
-
Response Surface Methodology (RSM): Fits a mathematical model (e.g., 2nd-order polynomial) to DOE results. Used to find optimum and understand parameter interactions.
-
Integration: Link simulation software (ANSYS, COMSOL) with optimization modules (e.g., ANSYS Optimizer) for shape/topology optimization.
3.6.3 User-Defined Functions (UDFs) & Customization
-
Purpose: Extend solver capabilities beyond built-in models.
-
Applications:
-
Custom boundary conditions (e.g., oscillating pressure, variable heat flux).
-
Custom source terms in governing equations.
-
Custom material properties (user-defined functions of temperature, strain rate).
-
Custom reports/force coefficients.
-
-
Implementation: Typically via C/C++/Fortran scripting compiled with the solver (e.g., ANSYS Fluent UDF, COMSOL Java).
[!TIP] UDFs are advanced and require programming knowledge. In lab exams, you may be asked to interpret a simple UDF code snippet (e.g., a time-dependent BC) rather than write one.