6.2 Fluctuating Stresses, Failure Criteria & Cumulative Damage

6.2 Fluctuating Stresses, Failure Criteria & Cumulative Damage

In practical mechanical systems, dynamic stresses rarely cycle symmetrically about a zero mean value. Connecting rods, turbine blades, pressurized pipes, and gear teeth experience fluctuating stresses characterized by combined steady (mean) loads and cyclic (alternating) loads. Evaluating components under non-zero mean stress requires specialized failure criteria plotted on the Haigh diagram, rigorous checks against initial yielding via the Langer line, and cumulative damage models for variable-amplitude spectrum loading.

---

Fluctuating Stress Components & Terminology

A general cyclic stress history fluctuating periodically between a maximum tensile stress $\sigma_{\max}$ and a minimum stress $\sigma_{\min}$ is decomposed into two distinct, orthogonal mathematical components:

   Stress ^
          |             _---_                       _---_
  sigma_max +---------/-------'\------------------/-------'\----------+
          |          /     |    \                /     |    \
          |         |      |     |              |      |     |        | 2 * sigma_a
  sigma_m + - - - - + - - -|- - -+ - - - - - - -+ - - -|- - -+ - - - -+ (Stress Range Delta sigma)
          |         |      |     |              |      |     |        |
          |          \     |    /                \     |    /
  sigma_min +---------\_______/------------------\_______/----------+
          |                |
          +----------------+------------------------------------------> Time
                           |<--- Period T --->|

Mathematical Definitions

1. Alternating Stress Amplitude ($\sigma_a$): The cyclic half-range that drives crack propagation:

$$\sigma_a = \frac{|\sigma_{\max} - \sigma_{\min}|}{2}$$

1. Mean Stress ($\sigma_m$): The static, time-invariant baseline stress level:

$$\sigma_m = \frac{\sigma_{\max} + \sigma_{\min}}{2}$$

1. Stress Range ($\Delta \sigma$): The full peak-to-peak excursion:

$$\Delta \sigma = \sigma_{\max} - \sigma_{\min} = 2\sigma_a$$

1. Stress Ratio ($R$): The ratio of minimum stress to maximum stress, serving as the universal descriptor of cyclic loading regime:

$$R = \frac{\sigma_{\min}}{\sigma_{\max}}$$

1. Amplitude Ratio ($A$): The ratio of alternating stress to mean stress:

$$A = \frac{\sigma_a}{\sigma_m} = \frac{\sigma_{\max} - \sigma_{\min}}{\sigma_{\max} + \sigma_{\min}} = \frac{1 - R}{1 + R}$$

Standard Benchmark Loading Regimes

Physical Impact of Mean Stress

• Tensile Mean Stress ($\sigma_m > 0$): Highly detrimental. Tensile mean stress pulls microcrack faces apart, facilitating Mode I crack opening, accelerating Stage II crack growth, and dramatically lowering the allowable alternating fatigue strength.
• Compressive Mean Stress ($\sigma_m < 0$): Highly beneficial. Compressive mean stress forces crack faces tightly shut, impeding crack propagation. In machine design, compressive mean stress is usually treated conservatively by ignoring the beneficial effect and setting $\sigma_m = 0$ (so $\sigma_a \le S_e$).

---

Fatigue Failure Criteria on the Haigh Diagram

The Haigh diagram plots alternating stress amplitude ($\sigma_a$) on the vertical axis against mean stress ($\sigma_m$) on the horizontal axis. Any cyclic stress state is represented as a discrete point $(\sigma_m, \sigma_a)$. When a component operates under proportional loading, the operating locus is a straight load line radiating from the origin with slope equal to the amplitude ratio:

$$r = \frac{\sigma_a}{\sigma_m} = A = \frac{1 - R}{1 + R}$$

  Alternating Stress (sigma_a) ^
                            |
                        S_e +-------------------------\
                            | \                        ---\
                            |  \                           ---\  Gerber Parabola
                            |   \  Modified Goodman            ---\
                            |    \                              ---\
                            |     \                                 ---\
                            |      \                                    ---\
                            +-------+-------------------+--------------------+---> Mean Stress (sigma_m)
                            0      S_y                 S_ut                 S_ut
                                    ^                   ^
                               Soderberg Line      Goodman & Gerber Intercept

Four classical failure theories define allowable operating boundaries in the first quadrant ($\sigma_m \ge 0, \sigma_a \ge 0$):

1. Modified Goodman Criterion

The Modified Goodman relation connects the fully corrected endurance limit $S_e$ on the $\sigma_a$-axis to the ultimate tensile strength $S_{ut}$ on the $\sigma_m$-axis via a straight line:

$$\frac{\sigma_a}{S_e} + \frac{\sigma_m}{S_{ut}} = \frac{1}{n_G}$$

Solving for the Goodman factor of safety $n_G$ along a constant load line ($r = \sigma_a / \sigma_m$):

$$n_G = \frac{1}{\frac{\sigma_a}{S_e} + \frac{\sigma_m}{S_{ut}}} = \frac{S_e S_{ut}}{\sigma_a S_{ut} + \sigma_m S_e}$$

2. Soderberg Criterion

The Soderberg line connects the endurance limit $S_e$ to the tensile yield strength $S_y$:

$$\frac{\sigma_a}{S_e} + \frac{\sigma_m}{S_y} = \frac{1}{n_S} \implies n_S = \frac{S_e S_y}{\sigma_a S_y + \sigma_m S_e}$$

The Soderberg criterion is overly conservative because it heavily penalizes mean stress by replacing $S_{ut}$ with $S_y$. While it guarantees that static yielding will never occur, it unnecessarily inflates component dimensions and weight.

3. Gerber Parabola

The Gerber criterion models the relationship as a parabolic curve connecting $S_e$ to $S_{ut}$:

$$\frac{n \sigma_a}{S_e} + \left(\frac{n \sigma_m}{S_{ut}}\right)^2 = 1$$

Rearranging into standard quadratic form $A n^2 + B n - 1 = 0$:

$$\left(\frac{\sigma_m}{S_{ut}}\right)^2 n^2 + \left(\frac{\sigma_a}{S_e}\right) n - 1 = 0$$

Applying the quadratic formula yields the Gerber factor of safety $n_{\text{Gerber}}$:

$$n_{\text{Gerber}} = \frac{-\frac{\sigma_a}{S_e} + \sqrt{\left(\frac{\sigma_a}{S_e}\right)^2 + 4\left(\frac{\sigma_m}{S_{ut}}\right)^2}}{2\left(\frac{\sigma_m}{S_{ut}}\right)^2} = \frac{1}{2}\left(\frac{S_{ut}}{\sigma_m}\right)^2 \left(\frac{\sigma_a}{S_e}\right) \left[-1 + \sqrt{1 + \left(\frac{2 \sigma_m S_e}{\sigma_a S_{ut}}\right)^2}\right]$$

The Gerber parabola accurately models the mean of experimental failure points for ductile alloys. However, because it represents an average fit, roughly 50% of real test specimens fail below the Gerber curve. Consequently, it must be paired with higher design factors of safety in mission-critical applications.

4. ASME Elliptic Criterion

The ASME Elliptic criterion connects $S_e$ and $S_y$ through a quarter-ellipse:

$$\left(\frac{n \sigma_a}{S_e}\right)^2 + \left(\frac{n \sigma_m}{S_y}\right)^2 = 1$$

Solving directly for the ASME Elliptic factor of safety $n_E$:

$$n_E = \frac{1}{\sqrt{\left(\frac{\sigma_a}{S_e}\right)^2 + \left(\frac{\sigma_m}{S_y}\right)^2}}$$

The ASME Elliptic curve is a smooth envelope that ends at $S_y$, but near the mean-stress axis it lies outside the Langer yield line, so the first-cycle yield check below is still required (Shigley pairs every fatigue criterion, including ASME Elliptic, with the Langer line).

---

The Langer Static Yield Line Check

Fatigue failure criteria dictate whether a component will develop propagating cracks over millions of cycles. However, fatigue criteria do not prevent gross plastic deformation or yielding during the very first load cycle. If the maximum peak stress $\sigma_{\max} = \sigma_m + \sigma_a$ exceeds the material's yield strength $S_y$, the part will undergo permanent structural deformation immediately upon startup.

  Alternating Stress (sigma_a) ^
                            |
                        S_y +------------\
                            |             ---\   Langer Static Yield Line
                        S_e +------\          ---\  (Slope = -1)
                            |       ---\          ---\ 
                            | Safe      ---\          ---\
                            | Operating     ---\          ---\
                            | Zone              ---\          ---\
                            +-----------------------+----------+---> Mean Stress (sigma_m)
                            0                      S_e        S_y

The Langer static yield line forms an absolute upper ceiling on the Haigh diagram connecting $(0, S_y)$ to $(S_y, 0)$ with a constant slope of $-1$:

$$\sigma_a + \sigma_m \le \frac{S_y}{n_y} \iff \sigma_{\max} \le \frac{S_y}{n_y}$$

Solving for the static yield safety factor $n_y$:

$$n_y = \frac{S_y}{\sigma_a + \sigma_m} = \frac{S_y}{\sigma_{\max}}$$

---

Multiaxial Fatigue & Fluctuating Stresses

Many machine elements experience combined fluctuating multiaxial stresses. A classic example is a power transmission shaft subjected to a completely reversed bending moment ($M$) combined with a steady drive torque ($T$), generating alternating normal bending stresses $\sigma_a$ and steady torsional shear stresses $\tau_m$.

Equivalent von Mises Alternating & Mean Stresses

For synchronous, in-phase cyclic multiaxial stresses, the alternating and mean stress states are independently converted into scalar von Mises equivalent stresses:

1. Equivalent Alternating von Mises Stress ($\sigma_a'$):

$$\sigma_a' = \sqrt{\sigma_a^2 + 3\tau_a^2}$$

1. Equivalent Mean von Mises Stress ($\sigma_m'$):

$$\sigma_m' = \sqrt{\sigma_m^2 + 3\tau_m^2}$$

Once $\sigma_a'$ and $\sigma_m'$ are computed, they are inserted directly into the uniaxial fatigue failure equations:

$$\frac{\sigma_a'}{S_e} + \frac{\sigma_m'}{S_{ut}} = \frac{1}{n_G} \quad \text{(Modified Goodman with von Mises)}$$ $$\sigma_a' + \sigma_m' \le \frac{S_y}{n_y} \quad \text{(Langer Yield with von Mises)}$$

Sines Method for Combined Cyclic Torsion & Bending

Extensive experimental testing shows that steady mean torsional shear stresses have virtually zero detrimental effect on the fatigue endurance limit of ductile metals. Sines formulated an alternative criterion that incorporates hydrostatic mean stress while excluding steady shear:

$$\frac{\sigma_a'}{S_e} + \frac{C_m (\sigma_{x,m} + \sigma_{y,m} + \sigma_{z,m})}{S_{ut}} \le \frac{1}{n}$$

Under pure steady torque $\tau_m$ (where normal mean stresses are zero: $\sigma_{x,m} = \sigma_{y,m} = 0$), the mean term vanishes completely, predicting that steady torque does not degrade alternating bending fatigue resistance.

---

Cumulative Fatigue Damage: The Palmgren-Miner Linear Damage Rule

Real-world mechanical systems operate under variable-amplitude spectrum loading rather than constant stress amplitudes. An automobile suspension encounters smooth asphalt, expansion joints, and occasional severe potholes; an aircraft wing experiences taxiing, climb gust turbulence, cruising, and landing impact.

  Stress ^      Block 1: n1 cycles at sigma_1           Block 2: n2 at sigma_2
         |       +----+   +----+   +----+                +--+  +--+  +--+
         |       |    |   |    |   |    |                |  |  |  |  |  |
         |       |    |   |    |   |    |  +--+  +--+    |  |  |  |  |  |
         +-------+----+---+----+---+----+--+--+--+--+----+--+--+--+--+--+---> Time
                                           Block 3: n3

The Palmgren-Miner Rule

The Palmgren-Miner linear damage hypothesis assumes that each individual load cycle at stress amplitude $\sigma_i$ consumes a constant fraction of the component's total fatigue life:

$$\text{Cycle Ratio for Block } i = \frac{n_i}{N_i}$$

where:

• $n_i$ is the number of cycles applied at stress amplitude $\sigma_i$.
• $N_i$ is the fatigue life (cycles to failure) if stress amplitude $\sigma_i$ operated alone, calculated from Basquin's equation $N_i = (\sigma_i / a)^{1/b}$.

The total cumulative damage fraction $D$ is the linear summation across all $k$ spectrum blocks:

$$D = \sum_{i=1}^k \frac{n_i}{N_i}$$

Failure is predicted when cumulative damage reaches unity:

$$D \ge 1.0$$

If the cumulative damage accrued over an operational inspection interval is $D$, the remaining useful life fraction is $(1 - D)$. If block $j$ continues until failure, the remaining allowable cycles $n_{\text{remaining}, j}$ is:

$$n_{\text{remaining}, j} = (1 - D) N_j$$

Critical Limitations & Load Sequencing Effects

While widely specified in design codes for its mathematical simplicity, Miner's rule possesses major engineering limitations:

1. Linearity & Commutativity Assumption: Miner's rule assumes damage accumulates linearly and independently of loading sequence ($A \to B$ accumulates identical damage to $B \to A$). In reality, fatigue damage is non-linear.
2. High-to-Low Sequencing ($H \to L$): When high-stress cycles occur early in service life, they rapidly initiate Stage I microcracks and generate adverse tensile residual stresses at notch roots. Subsequent low-stress cycles—which normally would require millions of cycles to nucleate a crack—immediately propagate the pre-existing crack. Under high-to-low loading, failure occurs at damage fractions significantly below unity:

$D_{\text{failure}} \approx 0.50\text{ to }0.80$

1. Low-to-High Sequencing ($L \to H$): Early cycles applied at low stress amplitudes can produce cyclic strain hardening (strain aging) and blunt microcrack tips (the coaxing effect). The component can endure cumulative damage fractions exceeding unity:

$D_{\text{failure}} \approx 1.20\text{ to }1.60$

1. Sub-Endurance Limit Cycling: In constant-amplitude testing, cycles below $S_e$ cause zero damage ($N = \infty$). In spectrum loading, once an early overload initiates a crack, subsequent stress cycles below the original endurance limit continue to propagate the crack, contributing actively to cumulative damage.

---

Comparison Table: Classical Fatigue Failure Criteria

---

Realistic Worked Engineering Scenario: Actuator Linkage Under Spectrum Loading

   P_min = 3,000 lbf                                  P_max = 15,000 lbf
   +===================[ Actuator Tie Rod ]===================+
   | Area A = 0.375 in^2 | Cold-Rolled AISI 1020 Steel       |
   | S_ut = 68.0 kpsi    | S_y = 57.0 kpsi                   |
   | Fully Modified Axial Endurance Limit: S_e = 22.0 kpsi   |
   | Fatigue Stress Concentration Factor:  K_f = 1.35        |
   +=========================================================+

Problem Statement

A structural tie rod in an automated packaging mechanism is subjected to a cyclic axial force fluctuating between $P_{\max} = 15,000\text{ lbf}$ (tension) and $P_{\min} = 3,000\text{ lbf}$ (tension). The rod is fabricated from cold-rolled AISI 1020 steel:

• Ultimate tensile strength: $S_{ut} = 68.0\text{ kpsi}$
• Yield strength: $S_y = 57.0\text{ kpsi}$
• Fully corrected axial endurance limit: $S_e = 22.0\text{ kpsi}$
• Critical cross-sectional area: $A = 0.375\text{ in}^2$
• Fatigue stress concentration factor at mounting eyelet: $K_f = 1.35$

Calculate:

1. Nominal and peak alternating and mean stresses ($\sigma_a, \sigma_m$), stress ratio $R$, and amplitude ratio $A$.
2. The first-cycle static yield factor of safety $n_y$ using the Langer line.
3. The fatigue factor of safety $n_G$ using the Modified Goodman criterion.
4. The fatigue factor of safety $n_{\text{Gerber}}$ using the Gerber parabola.
5. The cumulative damage fraction $D$ and remaining life if the rod subsequently experiences a three-block variable load spectrum:
   • Block 1: $n_1 = 30,000$ cycles at a stress level where $N_1 = 150,000$ cycles
   • Block 2: $n_2 = 120,000$ cycles at a stress level where $N_2 = 400,000$ cycles
   • Block 3: $n_3 = 500,000$ cycles at a stress level where $N_3 = 2,500,000$ cycles

---

Step-by-Step Mathematical Solution

Step 1: Stress Components & Ratios

1. Nominal maximum and minimum axial stresses:

$$\sigma_{\max,0} = \frac{P_{\max}}{A} = \frac{15,000\text{ lbf}}{0.375\text{ in}^2} = 40,000\text{ psi} = 40.0\text{ kpsi}$$ $$\sigma_{\min,0} = \frac{P_{\min}}{A} = \frac{3,000\text{ lbf}}{0.375\text{ in}^2} = 8,000\text{ psi} = 8.0\text{ kpsi}$$

1. Stress ratio $R$:

$$R = \frac{\sigma_{\min,0}}{\sigma_{\max,0}} = \frac{8.0\text{ kpsi}}{40.0\text{ kpsi}} = 0.200 \quad \text{(Tension--Tension Loading)}$$

1. Nominal alternating and mean stresses:

$$\sigma_{a,0} = \frac{\sigma_{\max,0} - \sigma_{\min,0}}{2} = \frac{40.0 - 8.0}{2} = 16.0\text{ kpsi}$$ $$\sigma_{m,0} = \frac{\sigma_{\max,0} + \sigma_{\min,0}}{2} = \frac{40.0 + 8.0}{2} = 24.0\text{ kpsi}$$

1. Amplitude ratio $A$:

$$A = \frac{\sigma_a}{\sigma_m} = \frac{16.0}{24.0} = 0.667$$

1. Actual peak stresses incorporating fatigue stress concentration $K_f = 1.35$:

$$\sigma_a = K_f \sigma_{a,0} = 1.35 \times 16.0\text{ kpsi} = 21.60\text{ kpsi}$$ $$\sigma_m = K_f \sigma_{m,0} = 1.35 \times 24.0\text{ kpsi} = 32.40\text{ kpsi}$$ $$\sigma_{\max} = \sigma_a + \sigma_m = 21.60 + 32.40 = 54.00\text{ kpsi}$$

Step 2: First-Cycle Yield Check (Langer Line)

Check whether peak transient stress causes plastic yield during initial application:

$$n_y = \frac{S_y}{\sigma_{\max}} = \frac{S_y}{\sigma_a + \sigma_m} = \frac{57.0\text{ kpsi}}{54.0\text{ kpsi}} = 1.056 \approx 1.06$$

Because $n_y = 1.06 > 1.0$, the tie rod will not yield on the first load application.

Step 3: Fatigue Factor of Safety: Modified Goodman

Applying the Modified Goodman equation:

$$\frac{\sigma_a}{S_e} + \frac{\sigma_m}{S_{ut}} = \frac{1}{n_G}$$ $$\frac{21.60\text{ kpsi}}{22.00\text{ kpsi}} + \frac{32.40\text{ kpsi}}{68.00\text{ kpsi}} = 0.98182 + 0.47647 = 1.45829$$ $$n_G = \frac{1}{1.45829} = 0.686 \approx 0.69$$

Because $n_G = 0.69 < 1.0$, the Modified Goodman criterion predicts fatigue failure prior to infinite life ($N < 10^6$ cycles).

Step 4: Fatigue Factor of Safety: Gerber Parabola

Applying the Gerber equation in quadratic form:

$$\left(\frac{\sigma_m}{S_{ut}}\right)^2 n^2 + \left(\frac{\sigma_a}{S_e}\right) n - 1 = 0$$

Compute coefficients:

$$\frac{\sigma_m}{S_{ut}} = \frac{32.40}{68.00} = 0.47647 \implies \left(\frac{\sigma_m}{S_{ut}}\right)^2 = 0.22702$$ $$\frac{\sigma_a}{S_e} = \frac{21.60}{22.00} = 0.98182$$

Quadratic equation: $0.22702 n^2 + 0.98182 n - 1 = 0$

$$n_{\text{Gerber}} = \frac{-0.98182 + \sqrt{(0.98182)^2 - 4(0.22702)(-1)}}{2(0.22702)} = \frac{-0.98182 + \sqrt{0.96397 + 0.90808}}{0.45404}$$ $$n_{\text{Gerber}} = \frac{-0.98182 + \sqrt{1.87205}}{0.45404} = \frac{-0.98182 + 1.36823}{0.45404} = \frac{0.38641}{0.45404} = 0.851 \approx 0.85$$

While Gerber is less conservative than Goodman ($0.85$ vs $0.69$), it also confirms $n < 1.0$, predicting finite life.

Step 5: Cumulative Damage Analysis (Palmgren-Miner)

Evaluating the three spectrum loading blocks:

$$D_1 = \frac{n_1}{N_1} = \frac{30,000}{150,000} = 0.200$$ $$D_2 = \frac{n_2}{N_2} = \frac{120,000}{400,000} = 0.300$$ $$D_3 = \frac{n_3}{N_3} = \frac{500,000}{2,500,000} = 0.200$$

Total cumulative damage fraction accrued:

$$D = D_1 + D_2 + D_3 = 0.200 + 0.300 + 0.200 = 0.700$$

1. Remaining Damage Capacity:

$$\text{Damage Margin} = 1.000 - D = 1.000 - 0.700 = 0.300 \quad (30\% \text{ remaining life})$$

1. Remaining Cycle Capacity Under Block 2 Conditions: If the linkage continues operating exclusively under the Block 2 loading regime:

$$n_{\text{remaining}, 2} = (1.000 - D) \times N_2 = 0.300 \times 400,000 = 120,000\text{ cycles}$$

1. Load Sequencing Caution: If Block 1 (the highest stress level) was applied first in the machine's life, early microcracking can cause failure at a damage sum well below 1.0 (values of about 0.5 to 0.8 are often reported). At $D = 0.70$, the tie rod is entering its critical wear-out phase and must be slated for immediate non-destructive ultrasonic inspection or scheduled replacement.