10.1 Static Failure Theories & Variable/Fatigue Loading

Static Failure Theories & Variable/Fatigue Loading

In mechanical engineering design, machine components are rarely subjected to simple uniaxial static tension. Heavy machinery used in open-cast and underground coal mining—such as draglines, shearers, haul trucks, continuous miners, and conveyor driveheads—endure complex multi-axial states of stress combined with severe cyclic, reversing, and shock loading. To prevent catastrophic failure, a mechanical design engineer must apply appropriate static failure theories for non-cyclic multi-axial stresses and fatigue failure models for fluctuating service loads.

---

1. Multi-Axial Static Failure Theories

A static failure theory correlates a multi-axial stress state $(\sigma_1, \sigma_2, \sigma_3)$ in a component to mechanical property limits determined from a standard uniaxial tensile test: the yield strength $(S_{yt})$ for ductile materials and the ultimate tensile strength $(S_{ut})$ for brittle materials.

1.1 Maximum Principal Stress Theory (Rankine's Theory / MPST)

• Statement: Failure occurs when the maximum principal tensile stress in a bi-axial or tri-axial stress system reaches the yield strength or ultimate tensile strength under simple tension. $\sigma_1 \ge \frac{S_{ut}}{N} \quad \text{or} \quad |\sigma_3| \ge \frac{S_{uc}}{N}$
• Failure Envelope: In the 2D principal stress plane $(\sigma_1, \sigma_2)$, the boundary is a square/rectangle defined by $\sigma_1 = \pm S_{ut}$ and $\sigma_2 = \pm S_{ut}$.
• Application: Valid only for brittle materials (e.g., cast iron, high-carbon hardened steels) where failure occurs by tensile cleavage across planes of maximum normal stress.
• Limitation: Totally unsuitable for ductile materials. Under pure shear $(\sigma_1 = \tau, \sigma_2 = -\tau)$, Rankine predicts shear yield strength $\tau_y = S_{yt}$, whereas ductile metals actually yield at $\tau_y \approx 0.50 S_{yt}$ to $0.577 S_{yt}$, creating an unsafe 73% to 100% overestimation of shear load capacity.

1.2 Maximum Shear Stress Theory (Tresca's / Guest's Theory / MSST)

• Statement: Yielding begins when the maximum shear stress in a multi-axial stress state equals the maximum shear stress at the onset of yielding in a standard uniaxial tension test. $\tau_{\max} = \max\left(\left|\frac{\sigma_1 - \sigma_2}{2}\right|, \left|\frac{\sigma_2 - \sigma_3}{2}\right|, \left|\frac{\sigma_3 - \sigma_1}{2}\right|\right) \le \frac{S_{yt}}{2N}$
• Pure Shear Prediction: For a state of pure shear $(\sigma_1 = \tau, \sigma_2 = -\tau, \sigma_3 = 0)$: $\tau_{\max} = \frac{\tau - (-\tau)}{2} = \tau \le \frac{S_{yt}}{2N} \implies S_{sy} = 0.5 S_{yt}$
• Failure Envelope: A regular hexagon oriented at $45^{\circ}$ within the $\sigma_1-\sigma_2$ coordinate system.
• Application & Characteristic: Standard conservative theory for ductile materials (e.g., mild steel, structural alloys). Because the hexagonal locus lies entirely within the Distortion Energy ellipse, Tresca always yields a conservative component dimension.

1.3 Distortion Energy Theory (Von Mises / Hencky Theory / DET)

• Statement: Yielding occurs when the distortion energy (shear strain energy causing shape distortion without volume change) per unit volume under multi-axial stress reaches the distortion strain energy at yield in simple uniaxial tension. $\text{Total Strain Energy Density } U = U_v (\text{volumetric}) + U_d (\text{distortion})$ $U_d = \frac{1+\nu}{6E}\left[(\sigma_1 - \sigma_2)^2 + (\sigma_2 - \sigma_3)^2 + (\sigma_3 - \sigma_1)^2\right]$
• Von Mises Equivalent Stress ($\sigma_v$ / $\sigma'$): $\sigma' = \sqrt{\frac{1}{2}\left[(\sigma_1 - \sigma_2)^2 + (\sigma_2 - \sigma_3)^2 + (\sigma_3 - \sigma_1)^2\right]} \le \frac{S_{yt}}{N}$
• For 2D plane stress states with $(\sigma_x, \sigma_y, \tau_{xy})$: $\sigma' = \sqrt{\sigma_x^2 - \sigma_x \sigma_y + \sigma_y^2 + 3\tau_{xy}^2} \le \frac{S_{yt}}{N}$
• Pure Shear Prediction: For pure shear $(\sigma_x = 0, \sigma_y = 0, \tau_{xy} = \tau)$: $\sigma' = \sqrt{3\tau^2} = \sqrt{3}\tau \le S_{yt} \implies S_{sy} = \frac{S_{yt}}{\sqrt{3}} \approx 0.577 S_{yt}$
• Failure Envelope: An ellipse inclined at $45^{\circ}$ with major axis along $\sigma_1 = \sigma_2$ and minor axis along $\sigma_1 = -\sigma_2$.
• Application: Most accurate failure theory for ductile materials. Predicts $15.5%$ higher allowable shear strength than Tresca $(0.577 / 0.500 = 1.155)$, allowing optimum weight and material efficiency.

1.4 Maximum Principal Strain Theory (St. Venant's Theory / MPET)

• Statement: Failure occurs when the maximum principal strain in a multi-axial stress state reaches the yield strain in uniaxial tension: $\epsilon_1 = \frac{\sigma_1}{E} - \nu\frac{\sigma_2 + \sigma_3}{E} \le \frac{S_{yt}}{E}$.
• Failure Envelope: A rhomboid / rhombus.
• Application: Historically used for thick-walled cylinders and gun barrels, but generally superseded by Von Mises and Tresca due to inconsistent experimental correlation.

1.5 Total Strain Energy Theory (Haigh's / Beltrami's Theory / TSET)

• Statement: Yielding occurs when the total strain energy per unit volume absorbs an amount equal to that stored at yield in simple tension: $U_{\text{total}} = \frac{1}{2E}\left[\sigma_1^2 + \sigma_2^2 + \sigma_3^2 - 2\nu(\sigma_1\sigma_2 + \sigma_2\sigma_3 + \sigma_3\sigma_1)\right] \le \frac{S_{yt}^2}{2E}$
• Failure Envelope: An ellipse with semi-axes distinct from Von Mises.
• Application: Suitable for ductile materials under hydrostatic compression/tension; over-predicts failure when volumetric dilation is large.

---

2. Stress Concentration & Notch Sensitivity

Geometric discontinuities (fillets, keyways, oil holes, shoulders, notches) disturb uniform stress trajectories, generating localized peak stresses far exceeding nominal cross-sectional values.

2.1 Theoretical Stress Concentration Factor $(K_t)$

$K_t = \frac{\sigma_{\max}}{\sigma_0} \quad \text{or} \quad K_{ts} = \frac{\tau_{\max}}{\tau_0}$ where $\sigma_{\max}$ is the true localized peak stress and $\sigma_0$ is the nominal stress calculated from elementary mechanics $(P/A, M/Z, T/Z_p)$ on the net cross-section. $K_t$ depends purely on component geometry and loading mode, independent of the material.

• Behavior under Static Loading:
  • In ductile materials, static stress concentration is generally neglected because localized plastic yielding at the notch root causes stress redistribution across the net section without macroscopic fracture.
  • In brittle materials, localized yielding cannot occur; stress concentration must be applied directly in static design: $\sigma_{\text{design}} = K_t \sigma_0$.

2.2 Fatigue Notch Factor $(K_f)$ and Notch Sensitivity $(q)$

Under cyclic fatigue loading, even ductile materials fail by progressive brittle crack growth initiated at notches. However, the actual fatigue degradation is frequently less severe than $K_t$. $K_f = \frac{\text{Endurance limit of unnotched specimen } (S_e)}{\text{Endurance limit of notched specimen } (S_{ef})}$ $\text{Notch Sensitivity Index } q = \frac{K_f - 1}{K_t - 1} \implies K_f = 1 + q(K_t - 1)$

• $q = 0 \implies K_f = 1$ (Material is completely insensitive to notches; typical for coarse cast iron).
• $q = 1 \implies K_f = K_t$ (Full theoretical notch sensitivity; typical for ultra-high-strength brittle alloy steels).
• Neuber's Equation: $q = \frac{1}{1 + \sqrt{a/r}}$, where $r$ is the notch root radius and $a$ is Neuber's material constant.

---

3. Variable Loading, Fatigue Mechanics & S-N Curves

Fatigue failure accounts for over $80%$ of all mechanical component breakdowns in mining haulage, gearboxes, and shafting. It occurs under dynamic repeating loads at stresses substantially below static yield strength.

3.1 Fatigue Mechanics & The Wöhler S-N Curve

Fatigue fracture evolves through three distinct physical phases: (1) Crack initiation at microscopic surface imperfections or slip bands; (2) Stage II striation propagation perpendicular to maximum tensile stress; (3) Catastrophic sudden fracture across the remaining net ligament.

• Low-Cycle Fatigue (LCF): Occurs at $N < 10^3$ cycles, characterized by significant macroscopic plastic deformation per cycle.
• High-Cycle Fatigue (HCF): Occurs at $N > 10^3$ to $10^6$ cycles, governed by elastic strain ranges.
• Endurance Limit ($S_e'$): For ferrous metals (steels, cast irons) and titanium alloys, the S-N curve displays a horizontal knee between $10^6$ and $10^7$ cycles. Below this stress amplitude, the specimen can withstand infinite cycles $(N = \infty)$ without failure.
  • For standard steels: $S_e' \approx 0.50 S_{ut}$ (for $S_{ut} \le 1400\text{ MPa}$); $S_e' = 700\text{ MPa}$ (for $S_{ut} > 1400\text{ MPa}$).
  • For cast iron: $S_e' \approx 0.40 S_{ut}$.
  • Non-ferrous alloys (aluminum, brass, copper) do not exhibit a distinct endurance limit; their fatigue strength is reported at a specified life (typically $S_{f}$ at $N = 5 \times 10^8$ cycles).

3.2 Marin Factors for Modified Endurance Limit

The laboratory endurance limit $S_e'$ is measured on a pristine, polished, unnotched R.R. Moore rotating-beam specimen ($d = 7.62\text{ mm}$) at room temperature. The actual component endurance limit $S_e$ is derived using Marin modification factors: $S_e = k_a \cdot k_b \cdot k_c \cdot k_d \cdot k_e \cdot k_f \cdot S_e'$

1. Surface Finish Factor $(k_a)$: Micro-roughness acts as crack initiation sites. $k_a = a \cdot S_{ut}^b$
   • Ground: $k_a \approx 0.90 - 0.95$
   • Machined / Cold-drawn: $k_a \approx 0.70 - 0.85$
   • Hot-rolled: $k_a \approx 0.50 - 0.70$
   • As-forged: $k_a \approx 0.30 - 0.50$
2. Size Factor $(k_b)$: Larger components have a higher statistical probability of flaws within the outer high-stress layer.
   • For bending/torsion: $k_b = 1.0$ for $d \le 7.62\text{ mm}$; $k_b = 1.24 d^{-0.107}$ for $7.62 < d \le 51\text{ mm}$; $k_b = 1.51 d^{-0.157}$ for $51 < d \le 250\text{ mm}$.
   • For axial loading: $k_b = 1.0$ (no gradient effect).
3. Load Factor $(k_c)$: Accounts for stress state difference relative to standard rotating beam bending.
   • Bending: $k_c = 1.0$
   • Axial loading: $k_c = 0.85$
   • Torsion / Pure shear: $k_c = 0.59$ (consistent with Von Mises $1/\sqrt{3}$)
4. Temperature Factor $(k_d)$: Accounts for creep and reduction in tensile strength at elevated operating temperatures ($k_d = 1.0$ for $T \le 350^{\circ}\text{C}$).
5. Reliability Factor $(k_e)$: Standard S-N curves represent $50%$ survival probability ($z_a = 0, k_e = 1.0$). For $90%$ reliability: $k_e = 0.897$; for $99%$ reliability: $k_e = 0.814$; for $99.9%$ reliability: $k_e = 0.753$.
6. Miscellaneous Effects Factor $(k_f)$: Includes corrosion pitting ($k_f \approx 0.3 - 0.5$), fretting fatigue, surface coatings (galvanizing/electroplating), and residual stresses (shot peening increases $k_f > 1.0$).

---

4. Fluctuating Stress Criteria & Fatigue Design Envelopes

When a machine member experiences cyclic loads where stress oscillates between a maximum value $\sigma_{\max}$ and a minimum value $\sigma_{\min}$:

• Mean Stress (Static component): $\sigma_m = \frac{\sigma_{\max} + \sigma_{\min}}{2}$
• Alternating Stress Amplitude (Dynamic component): $\sigma_a = \frac{\sigma_{\max} - \sigma_{\min}}{2}$
• Stress Ratio: $R = \frac{\sigma_{\min}}{\sigma_{\max}}$
  • Completely reversed: $R = -1, \sigma_m = 0$
  • Repeated loading (zero-to-peak): $R = 0, \sigma_m = \sigma_a$
  • Fluctuating tensile: $0 < R < 1$
• Amplitude Ratio: $A = \frac{\sigma_a}{\sigma_m} = \frac{1-R}{1+R}$

4.1 Classical Design Equations on the $\sigma_m - \sigma_a$ Diagram

     Alternating Stress (sigma_a)
     ^
 Se  + . . . . . . . . . . . . . . . .
     |\   Gerber Parabola             .
     | \.  (Passes through Se & Sut)   .
     |  \ '.                           .
     |   \   '.   Modified Goodman     .
     |    \    '. (Passes through Se & Sut)
     |     \     '.                    .
     |      \      '.  Soderberg Line  .
     |       \       '.(Passes through .
     |        \        '. Se & Syt)    .
     |         \         '.            .
     +----------+----------+-----------+---> Mean Stress (sigma_m)
     0         Syt        Sut

1. Soderberg Line (Yield-Based / Most Conservative): Connects endurance limit $S_e$ on the alternating axis to yield strength $S_{yt}$ on the mean axis. It eliminates any possibility of yielding during startup cycles. $\frac{\sigma_m}{S_{yt}} + \frac{\sigma_a}{S_e} = \frac{1}{N}$
2. Modified Goodman Line (Ultimate-Based / Standard Industrial Design): Connects endurance limit $S_e$ to ultimate tensile strength $S_{ut}$. Standard design criterion for ductile metals in mechanical engineering. $\frac{\sigma_m}{S_{ut}} + \frac{\sigma_a}{S_e} = \frac{1}{N}$
3. Gerber Parabola (Mean Experimental Fit): Represents parabolic regression curve passing through mean test fracture data. $\left(\frac{N \sigma_m}{S_{ut}}\right)^2 + \frac{N \sigma_a}{S_e} = 1 \implies \frac{\sigma_a}{S_e} = \frac{1}{N}\left[1 - \left(\frac{N \sigma_m}{S_{ut}}\right)^2\right]$
4. ASME Elliptic Line: Combines yield criteria with endurance limits in an elliptic envelope: $\left(\frac{N \sigma_m}{S_{yt}}\right)^2 + \left(\frac{N \sigma_a}{S_e}\right)^2 = 1$
5. Langer Static Yield Line (Yield Check for Goodman/Gerber): Because the Goodman line extends to $S_{ut}$, the peak stress $\sigma_{\max} = \sigma_m + \sigma_a$ can exceed yield strength $S_{yt}$, producing static plastic deformation on the very first cycle. Therefore, the design must be capped by the Langer line: $\frac{\sigma_m + \sigma_a}{S_{yt}} = \frac{1}{N} \implies \sigma_{\max} \le \frac{S_{yt}}{N}$

4.2 Comprehensive Comparison Table

---

5. Step-by-Step Worked Numerical Example

Problem: A transmission tie-rod forged from structural alloy steel has $S_{ut} = 600\text{ MPa}$, $S_{yt} = 400\text{ MPa}$, and a corrected endurance limit $S_e = 200\text{ MPa}$. It is subjected to a cyclic tensile force oscillating between $P_{\min} = 20\text{ kN}$ and $P_{\max} = 100\text{ kN}$. Determine the required rod diameter $d$ for infinite life using (a) Soderberg criterion and (b) Modified Goodman criterion with a factor of safety $N = 2.0$.

Solution:

1. Mean and Alternating Loads: $P_m = \frac{100 + 20}{2} = 60\text{ kN} = 60,000\text{ N}$ $P_a = \frac{100 - 20}{2} = 40\text{ kN} = 40,000\text{ N}$ $\sigma_m = \frac{P_m}{A} = \frac{60000}{\frac{\pi}{4} d^2} = \frac{76394}{d^2}\text{ MPa}, \quad \sigma_a = \frac{P_a}{A} = \frac{40000}{\frac{\pi}{4} d^2} = \frac{50930}{d^2}\text{ MPa}$
2. (a) Soderberg Design: $\frac{\sigma_m}{S_{yt}} + \frac{\sigma_a}{S_e} = \frac{1}{N} \implies \frac{76394}{400 d^2} + \frac{50930}{200 d^2} = \frac{1}{2.0}$ $\frac{190.99 + 254.65}{d^2} = 0.50 \implies \frac{445.64}{d^2} = 0.50 \implies d^2 = 891.28 \implies d = 29.85\text{ mm}$
3. (b) Modified Goodman Design: $\frac{\sigma_m}{S_{ut}} + \frac{\sigma_a}{S_e} = \frac{1}{N} \implies \frac{76394}{600 d^2} + \frac{50930}{200 d^2} = \frac{1}{2.0}$ $\frac{127.32 + 254.65}{d^2} = 0.50 \implies \frac{381.97}{d^2} = 0.50 \implies d^2 = 763.94 \implies d = 27.64\text{ mm}$
4. Static Yield (Langer) Verification for Goodman: $\sigma_{\max} = \sigma_m + \sigma_a = \frac{76394 + 50930}{(27.64)^2} = \frac{127324}{763.97} = 166.66\text{ MPa}$ $\text{Allowable Yield Stress } = \frac{S_{yt}}{N} = \frac{400}{2} = 200\text{ MPa} > 166.66\text{ MPa} \quad (\text{Safe against initial yield})$