FEA WORKFLOW · MECHANICAL ENGINEERING ANSYS · WB
Analyzing a Part
Model Using Ansys
Workbench
A practical finite element workflow — from CAD import to
validated stress results.
STATIC STRUCTURAL · LINEAR FEA · 10 SLIDES
Workshop Dark Edition · v2024 FIG. 0 — HERO / FEA BRACKET CONTOUR ● 01 / 10
Five linked stages from CAD to validated
WORKFLOW AT A GLANCE
results
Ansys Workbench chains each step as a project cell — outputs of one feed the next, so a geometry or material
change automatically propagates through to results. 02 / 09
01 02 03 04 05
Geometry & Meshing Boundary Solve Results
Materials Conditions & Loads
Discretize into finite Configure solver, run, Inspect stress, deformation,
Import CAD, simplify, assign elements; control size, type, Apply fixtures, contacts, monitor Newton–Raphson factor of safety; validate.
linear elastic material quality. forces, pressure, gravity. convergence.
properties.
AT A GLANCE
5 LINKED STAGES 2– TYPICAL BUILD 1 PARAMETERIZED MODEL N1.8 SOLVE-TIME SCALING
end-to-end chain
4h single source of truth node-count exponent
TIME
CAD → first run
STEP 01 / 05 GEOMETRY & MATERIALS
Geometry & Materials
Import the part, clean the load path, then assign an isotropic linear elastic material
from Ansys Engineering Data.
Import: STEP / IGES / Parasolid from SolidWorks, Creo, NX, or SpaceClaim — preserve mm-
kg-s units.
Simplify: defeature fillets <0.5 mm, embossed logos, tapped holes that do not affect the
load path.
Material: Structural Steel from Engineering Data — E = 200 GPa, ν = 0.30, ρ = 7850 kg/m³,
σ_yield = 250 MPa.
Coordinate check: align primary load axis with global X so BCs read cleanly downstream.
Volume check: confirm mass vs. CAD mass properties — divergence > 2% means geometry
import dropped a face.
Tip: keep a named DesignXplorer parameter on the load magnitude so the model
lightbulb can sweep later.
Fig. 1 — Bracket CAD model ready for FEA pre- 03 /
processing. 10
STEP 02 / 05 MESHING
Meshing
Discretize the part into finite elements. Mesh density drives accuracy, but it also
drives solve time — choose deliberately, not by default.
Element type: 10-node tetrahedra (SOLID187) for general geometry; hex-dominant for
swept / extruded bodies.
Sizing: global relevance 100; body sizing 2 mm on fillets and contact zones.
Refinement: 3-level refinement at high-stress regions using Body Sizing + Face Meshing.
Quality targets: Avg Skewness < 0.45, Orthogonal Quality > 0.50, Max Aspect Ratio < 20.
Count: typical bracket 50k–250k nodes; solve time scales roughly as N1.8.
187 <0.45 >0.50
Fig. 2 — Tetrahedral vs. hexahedral element mesh on 04 /
NODES / ELM SKEW ORTHO
the part. 10
STEP 03 / 05 BOUNDARY CONDITIONS & LOADS
Boundary Conditions & Loads
Constrain the part to remove rigid-body motion, then apply the real-world loads.
Wrong BCs invalidate everything downstream.
Fixed Support: clamp two bolt holes (all 6 DOF zeroed).
Cylindrical Support: pin bore — radial fixed, axial and tangential free.
Force: 5 kN applied normal to the load face, distributed by area.
Pressure: 2.5 MPa on the contact pad.
Standard Earth Gravity: 9.81 m/s² along −Y.
Contacts: bonded (default) for assembly; frictional μ = 0.15 where slip is expected.
Check: Σ reaction forces at supports must equal Σ applied loads — otherwise
balance the model is unbalanced.
Fig. 3 — Fixtures and applied loads on the bracket. 05 / 10
Solver Setup & Run
STEP 04 / 05 SOLVER · NEWTON–RAPHSON CONVERGENCE
Pick the right solver type, enable weak springs to guard against rigid-body drift, then watch the Newton–Raphson residual converge.
tune memory lock
Analysis Settings Solver Type Weak Springs
Static Structural, Large Deflection OFF (linear). 1 Direct (sparse) for <200k DOF; Iterative PCG for ON — suppresses rigid-body drift when BCs under-
substep, auto time-stepping OFF for linear static. larger models. Workbench auto-selects based on constrain. Disable for final run once stability is
node count. confirmed.
shutter_speed show_chart schedule
Inertia Relief Convergence Run Time
OFF — only use when no real fixture exists. Auto- Force-based residual < 0.5%. Newton–Raphson ~3–8 min for a 100k-node model on an 8-core
balances applied loads for free-body analysis. iterations typically converge in 3–5 steps for linear workstation, 32 GB RAM. Solve time ≈ k · N1.8.
elastic.
1 SOLVER 6 DOF/NODE 4 NR ITERATIONS 0.5% RESIDUAL
Results & Post-Processing
STEP 05 / 05 RESULTS · POST-PROCESSING
Extract the engineering answers — peak stress, deflection, and factor of safety — and sanity-
check them against hand calcs before signing off.
Equivalent (von Mises) Stress: peak 178 MPa on the inner fillet — below σ_yield of 250 MPa.
Total Deformation: max 0.42 mm at the load point; design limit is 0.50 mm.
Directional Deformation: load-axis deflection matches hand-calc Mc/I within 8%.
Factor of Safety: 1.40 minimum across the model; recommend > 1.5 for ductile steel under static
load.
Probe & verify: reaction-force sum at supports = applied load ±0.3% — model is balanced.
178 MPa 0.42 mm 1.40
PEAK STRESS DEFLECTION FACTOR OF SAFETY
Fig. 4 — von Mises stress contour on the part (Ansys).
Mesh Convergence & Validation
VERIFICATION
Peak stress is mesh-dependent until it stabilizes. Run a refinement study and compare against a hand calc before trusting any FEA number.
MESH DENSITY ELEMENTS PEAK STRESS (MPA) SOLVE TIME
Coarse 8,400 142 0:42
Medium 52,300 167 2:15
Fine REC OMME N DE D 198,000 176 7:48
Very Fine 612,000 178 24:10
Convergence Validation
Peak stress stabilizes at ~178 MPa beyond 200k elements. The medium mesh (52k) Hand calc σ = Mc/I gives 165 MPa at the fillet. FEA peak (178 MPa) sits within 8% of
under-predicts by ~6% — acceptable for design iteration, not for sign-off. the hand calc — the stress concentration at the fillet root explains the small excess.
Acceptable.
RULE: NEVER SIGN OFF ON PEAK STRESS WITHOUT A CONVERGENCE STUDY.
Five stages,
CLOSING
one repeatable workflow.
Geometry → Mesh → Boundary Conditions → Solve → Results. Master the chain, then drive the
model with parameters and convergence studies.
Workflow is repeatable: parameterize loads and mesh size so design iterations take minutes, not
hours.
Mesh matters: a convergence study is mandatory before signing off on peak stress values.
Validate: always compare FEA against hand calcs or test data; never trust a single run.
Speed vs. accuracy: medium mesh for design iterations, fine mesh for final sign-off.
Document: keep material spec, BC rationale, and mesh stats alongside every results report.
Ansys Workbench · Static Structural · Workshop Dark Edition Fig. 5 — Deformation field, final validated run.
09 / 10
ANSYS · WB · END OF DECK
Thank You.
Questions, mesh convergence deep-dives, or a walkthrough of the bracket
model — happy to dig in.
ANSYS WORKBENCH · STATIC STRUCTURAL · WORKSHOP DARK EDITION
Workshop Dark Edition · v2024 10 / 10