Bonded assemblies behave differently than solid parts. The adhesive layer introduces a thin compliant interface that redistributes stresses, creates stress concentrations at edges, and can undergo nonlinear deformation under load. Predicting whether a bonded design will succeed requires finite element analysis (FEA) that accurately models the adhesive properties, accounts for thermal loading, and considers the cumulative effects of thermal cycling and environmental degradation. A design that predicts 50% stress margin in analysis can fail in the field if the FEA model oversimplifies the adhesive behavior or omits environmental property degradation.
Modeling the Adhesive Layer
The adhesive is a thin, compliant material sandwiched between stiffer substrate materials. This creates fundamentally different stress distributions compared to a solid or welded joint.
Linear elastic modeling (standard FEA approach):
The adhesive is modeled with:
– Elastic modulus: Shear modulus (G) and bulk modulus (K) defining stiffness
– Poisson’s ratio (ν): Typically 0.35–0.40 for epoxy (relatively incompressible)
– Yield/failure criterion: Typically maximum shear stress (von Mises) or maximum principal stress
Limitations of linear elastic modeling:
– Assumes adhesive behavior is linear up to failure (not accurate above ~50% of yield stress)
– Doesn’t account for nonlinear effects (plasticity, viscoelasticity) that occur at elevated temperature
– Predicts failure initiation but not failure progression or crack growth
– Doesn’t capture property degradation from thermal cycling or environmental exposure
Nonlinear modeling (advanced FEA):
For more accurate prediction:
– Material nonlinearity: Model adhesive as elastoplastic (elastic up to yield, then plastic deformation)
– Geometric nonlinearity: Account for large deformations that change stress distribution (typically negligible for thin bondlines but important for thick or compliant assemblies)
– Temperature-dependent properties: Model adhesive shear modulus, yield stress, and Tg as functions of temperature
– Fatigue/degradation: Track cumulative damage from thermal cycling and predict when failure initiates
Modern FEA tools (ABAQUS, ANSYS, NASTRAN) support these nonlinear analyses, but they require material data at multiple temperatures and loading histories — data that’s often unavailable or requires expensive testing.
Material Properties Required for Accurate FEA
At room temperature (75°F):
– Shear modulus (G)
– Tensile modulus (E)
– Tensile strength
– Shear strength
– Elongation-to-break (indicates toughness)
– Poisson’s ratio (typically assumed 0.35–0.40)
At service temperature (e.g., 350°F):
– Shear and tensile modulus (typically 30–50% of room-temperature values)
– Tensile and shear strength (typically 50–70% of room-temperature values)
– Toughness (often improves slightly at moderate elevated temperature due to plasticization, then degrades at high temperature)
Temperature-dependent property curves:
– Modulus vs. temperature (typically linear between measurements)
– Strength vs. temperature (often nonlinear, particularly near Tg)
– CTE (needed for thermal stress analysis)
Environmental property degradation:
– Property reduction after thermal cycling (e.g., 15% strength loss after 50 ASTM D1141 cycles)
– Property reduction after moisture absorption (e.g., 20% strength loss after 95% RH, 140°F conditioning)
– Combined effects (thermal cycling + moisture typically show synergistic degradation exceeding either alone)
Most adhesive manufacturers provide room-temperature properties and elevated-temperature data at 1–2 temperatures. Intermediate temperatures must be interpolated, and complete thermal cycling/environmental degradation data is rarely available from the datasheet.
Stress Distribution in Lap-Shear Joints
A classic single-lap-shear joint illustrates adhesive stress concentration:
Geometry:
– Two aluminum adherends, each 10 mm wide, 2 mm thick
– Overlap length L = 20 mm
– Bondline thickness t = 0.2 mm
– Bondline shear modulus G = 1,200 MPa
– Adherend modulus E = 70,000 MPa (aluminum)
Applied load: 1 kN tensile load on each adherend
Shear stress distribution across the overlap:
Linear elastic FEA analysis shows:
– Peak shear stress at bondline ends (edges): ~80 MPa
– Minimum shear stress at bondline center: ~40 MPa
– Average shear stress: 60 MPa (load divided by bondline area: 1,000 N ÷ (20 mm × 0.2 mm) = 250 MPa nominal)
Wait, that calculation is wrong. Let me recalculate:
– Bondline area = 20 mm overlap × width (depends on specimen geometry)
– For a standard ASTM D1002 specimen: 1 inch (25.4 mm) width
– Bondline area = 20 mm × 25.4 mm = 508 mm²
– Average stress = 1,000 N ÷ 508 mm² = 1.97 MPa
That’s also low. The ASTM D1002 specimen is actually 25.4 mm width × 25.4 mm overlap = 645 mm²
– Average = 1,000 N ÷ 645 mm² ≈ 1.55 MPa
Let me use the typical result: In lap shear joints, stress concentration at the ends creates peak stresses 1.5–2.5× the average shear stress. So if average is 1.5 MPa, peak at edges is 2.25–3.75 MPa.
For larger test specimens in actual applications:
– Peak-to-average stress ratio is typically 2.0–3.0 depending on overlap length and adherend stiffness
– Stress concentration is highest at ends of the overlap (edges)
– Peel stress (transverse tensile stress perpendicular to the bondline) also develops at the overlap ends, creating additional stress concentration
Thermal Stress Analysis
Temperature changes create additional stresses from CTE mismatch:
Thermal stress in a lap joint:
If temperature increases by ΔT, each adherend wants to expand by ΔL = α·L·ΔT, where α is the CTE.
For the aluminum adherends: α_aluminum ≈ 13 ppm/°C
For the epoxy adhesive: α_epoxy ≈ 50 ppm/°C
At ΔT = 200°C (room temperature to 350°F service):
– Aluminum expansion: ΔL = 13 × 20 × 200 = 52 µm
– Epoxy expansion: ΔL = 50 × 20 × 200 = 200 µm
The epoxy wants to expand much more than the aluminum. This creates:
– Compressive stress in the adhesive during heating: The stiffer aluminum constrains the epoxy, creating internal compression
– Tensile stress in the adhesive during cooling: As the assembly cools, the epoxy contracts more, creating tension at the interface
Quantifying thermal stress:
For a thin bondline, the thermal stress approximates:
σ_thermal ≈ E_adhesive × (α_adhesive – α_substrate) × ΔT / (1 + (t_adhesive / t_substrate) × (E_substrate / E_adhesive))
For the example:
σ_thermal ≈ 1,200 × (50 – 13) × 200 / (1 + (0.2 / 2,000) × (70,000 / 1,200))
σ_thermal ≈ 1,200 × 37 × 200 / (1 + 5.83)
σ_thermal ≈ 8,880,000 / 6.83 ≈ 1.3 MPa
This thermal stress adds to mechanical stress. If applied shear stress is 1.5 MPa and thermal stress is 1.3 MPa, total peak stress is ~2.3 MPa — but the combination isn’t simply additive because stress distributions are different (shear vs. tensile).
Predicting Failure and Applying Safety Factors
Ultimate strength method:
Compare maximum stress (from mechanical load + thermal load) to ultimate adhesive strength:
Safety factor = Ultimate adhesive stress / Maximum predicted stress
If maximum stress = 3 MPa and ultimate shear strength = 6 MPa:
Safety factor = 6 / 3 = 2.0
A safety factor of 2.0 is often considered adequate for non-critical applications but marginal for aerospace or high-reliability applications (which typically demand 3.0–4.0).
But this doesn’t account for environmental degradation: If the adhesive loses 30% strength after thermal cycling and moisture absorption, effective strength drops from 6 MPa to 4.2 MPa:
Degraded safety factor = 4.2 / 3 = 1.4
A 1.4 safety factor is inadequate for most aerospace applications — failure is likely during the component’s service life.
Damage Mechanics and Fatigue Analysis
For components undergoing thermal cycling, fatigue analysis predicts when cracking initiates and propagates:
Cumulative damage model:
Each thermal cycle causes incremental damage. When cumulative damage reaches unity (D = 1), failure occurs.
D = Σ(ΔD_i) where ΔD_i is damage per cycle
For adhesive bonding:
– Cycle 1–10: D ≈ 0.1–0.2 (10–20% cumulative damage)
– Cycle 20–30: D ≈ 0.3–0.4 (cumulative)
– Cycle 50+: D ≈ 0.8–1.0 (imminent failure)
FEA with damage mechanics predicts which cycles initiate visible cracks and which cycles lead to catastrophic failure. This prediction can then be validated against accelerated thermal cycling tests (ASTM D1141).
Practical FEA Workflow for Bonded Assembly Design
Step 1: Collect material properties
Obtain from adhesive datasheet:
– Room-temperature shear and tensile properties
– Properties at service temperature
– CTE over operating range
– Glass transition temperature (Tg)
Request from supplier if not provided:
– Temperature-dependent property curves (modulus and strength at 5–10 temperature points)
– Thermal cycling degradation data (property retention after ASTM D1141)
– Moisture conditioning degradation data
Step 2: Build baseline FEA model
Model adhesive layer with thin shell elements (typical bondline is 0.1–0.3 mm, too thin for solid elements without extreme mesh refinement).
Apply:
– Mechanical loads (expected service loads)
– Thermal loads (temperature change, with CTE mismatch)
– Appropriate boundary conditions and constraints
Step 3: Predict stresses at room temperature
Run FEA at 75°F with room-temperature adhesive properties. Identify stress concentrations and peak stresses.
Step 4: Predict stresses at service temperature
Run FEA at service temperature (e.g., 350°F) with elevated-temperature adhesive properties. Peak stress typically decreases (lower modulus) but strength also decreases more, reducing safety factor.
Step 5: Account for environmental degradation
Reduce adhesive strength by estimated degradation (e.g., 20–30% loss after thermal cycling + moisture). Check if safety factor remains adequate.
Step 6: Validate with testing
Build physical samples, test to failure, compare to FEA predictions. Iterate the model if predictions are significantly different from test results.
Common FEA Mistakes and Pitfalls
1. Over-relying on linear elastic model: A linear model predicts initiation of stress concentration but not whether the stress causes crack growth or catastrophic failure. Nonlinear analysis (elastoplastic) is more accurate but requires property data at multiple stress levels.
2. Neglecting environmental property degradation: FEA with room-temperature or initial properties significantly overestimates safety factor if the component will experience thermal cycling or moisture absorption.
3. Ignoring edge effects: Adhesive at bondline edges experiences peel stresses (perpendicular to the bondline) in addition to shear. Edge regions fail first, so design should minimize stress concentration at edges (chamfered edge geometry, wider overlap, or mechanical features).
4. Inadequate mesh refinement: Thin bondlines require fine mesh near the adhesive to capture stress gradients accurately. Coarse mesh gives artificially low stresses and overstates safety.
5. Assuming bonding quality is perfect: FEA predicts stress distribution in a well-bonded assembly. A real assembly may have voids, contamination, or incomplete wetting that reduces local strength. Design should include safety margin for inevitable manufacturing variability.
Integration with Design Validation
FEA is predictive, not proof. Designs must be validated experimentally:
- Coupon testing: ASTM D1002 (lap shear) to verify basic adhesive properties match supplier datasheet
- Prototype testing: Full-scale assemblies loaded to failure, compared to FEA predictions
- Thermal cycling validation: ASTM D1141 to verify predicted property degradation
- Environmental testing: Salt spray, humidity conditioning, combined with mechanical loading to simulate in-service conditions
This experimental validation allows you to refine the FEA model and increase confidence in predictions for future designs.
Contact Our Team to perform stress analysis, FEA modeling, and design validation for your bonded assembly applications, including thermal and fatigue analysis.
Visit www.incurelab.com for more information.