1. Why We Need Failure Theories

Material properties (yield strength, ultimate strength) are measured in simple uniaxial tension or compression tests. But in real engineering, components experience multi-axial stress states — combinations of normal stresses and shear stresses acting simultaneously (e.g., a shaft under combined bending and torsion).

The question failure theories answer is: given a multi-axial stress state (σ₁, σ₂, σ₃) and the known uniaxial yield strength σ_y, will the material fail?

Each theory proposes a different criterion for what “causes” failure — maximum stress, maximum shear, maximum strain, or maximum distortion energy. No single theory works perfectly for all materials, but decades of experimental validation have shown which theories work best for which material types.

2. Maximum Normal Stress Theory (Rankine)

Best for: Brittle materials (cast iron, concrete, glass, ceramics) under static loading.

Limitation: Ignores the effect of the other principal stresses. Fails to predict yielding of ductile materials under combined loading accurately.

Failure envelope: Square on the σ₁-σ₂ plane.

3. Maximum Shear Stress Theory (Tresca)

In 2D (σ₃ = 0), if σ₁ and σ₂ have the same sign: max(|σ₁|, |σ₂|) ≤ σ_y

If σ₁ and σ₂ have opposite signs: |σ₁ − σ₂| ≤ σ_y

Best for: Ductile materials. Simple to apply. Slightly conservative (predicts failure before it actually occurs, giving a built-in safety margin).

Failure envelope: Hexagon on the σ₁-σ₂ plane, inscribed within the Von Mises ellipse.

4. Distortion Energy Theory (Von Mises)

For 2D stress (σ₃ = 0):

Best for: Ductile materials — the most accurate and widely used theory. Matches experimental data for ductile metals within 5%.

Failure envelope: Ellipse on the σ₁-σ₂ plane, circumscribing the Tresca hexagon.

5. Maximum Strain Theory (Saint-Venant)

Limited use: This theory is not widely used because it does not match experimental data well for most materials. It is included for completeness and because it occasionally appears in exam questions.

6. Side-by-Side Comparison

7. Which Theory to Use?

Key rule: The choice depends on the material type, not the loading type. Ductile materials use Von Mises/Tresca regardless of whether the loading is bending, torsion, combined, or anything else.

8. Worked Numerical Examples

Example 1: Von Mises — Biaxial Stress

Problem: A ductile steel component (σ_y = 250 MPa) has principal stresses σ₁ = 180 MPa, σ₂ = −60 MPa. Is it safe according to Von Mises?

Example 2: Tresca — Same Problem

Example 3: Pure Shear — Yield Prediction

Problem: A shaft made of steel (σ_y = 300 MPa) is under pure torsion. Find the shear stress at yielding using both Tresca and Von Mises.

9. Common Mistakes Students Make

• Using Von Mises for brittle materials: Von Mises and Tresca predict yielding — brittle materials don’t yield, they fracture. Use Maximum Normal Stress or Mohr’s theory for brittle materials.
• Confusing Tresca and Von Mises in pure shear: Tresca gives τ_y = 0.5σ_y; Von Mises gives τ_y = 0.577σ_y. The difference is 15% — this matters in design.
• Forgetting to consider the third principal stress: In 2D problems where σ₃ = 0, students sometimes forget that σ₃ = 0 is still a principal stress. For Tresca, τ_max = (σ_max − σ_min)/2, and σ_min might be σ₃ = 0, not σ₂.
• Not computing principal stresses first: Von Mises and Tresca criteria are expressed in terms of principal stresses. If given σ_x, σ_y, τ_xy, you must first find σ₁ and σ₂ using Mohr’s circle before applying the failure criterion.

10. Frequently Asked Questions

What is the Von Mises failure theory?

The Von Mises (Distortion Energy) theory states that yielding in ductile materials occurs when the distortion energy per unit volume reaches the value at yield in a simple tension test. It predicts failure when σ_VM = √[(σ₁−σ₂)² + (σ₂−σ₃)² + (σ₃−σ₁)²]/√2 ≥ σ_y. It is the most widely used and experimentally validated theory for ductile materials.

When should I use Tresca vs Von Mises?

Both are for ductile materials. Von Mises is more accurate (matches experiments within 5%). Tresca is simpler to compute and more conservative (predicts failure about 15% earlier than Von Mises). In practice, use Von Mises for design and analysis. Use Tresca when a quick, conservative estimate is acceptable or when the problem specifically asks for it.