Tools › Structures › Fatigue › Stress-Life Fatigue and Goodman Diagram


Stress-Life Fatigue and Goodman Diagram

Enter a constant-amplitude stress cycle and the ultimate, yield and endurance strengths of the material; get the factor of safety for infinite life by the modified Goodman, Soderberg, Gerber and ASME-elliptic criteria, the first-cycle yield check, and, when infinite life is not reached, the cycles to failure from the S-N line, with your cycle drawn on the fatigue diagram. A textbook estimate for smooth, constant-amplitude loading: not a certification analysis.

  • Alternating and mean stress, stress ratio R and amplitude ratio A from the maximum and minimum stress

  • Factor of safety by modified Goodman, Soderberg, Gerber and ASME-elliptic, and against first-cycle yield, each as formula, numbers, answer

  • Finite life: equivalent fully reversed stress (Goodman, Morrow, Walker / Smith-Watson-Topper) and cycles from S = a Nb

  • The fatigue diagram with every criterion line, your operating point and its load line, and the S-N line on log axes

Stress-Life Fatigue


Unit System: /

/

1. Stress Cycle






2. Material Strengths







3. Endurance Limit Factors (1 = no change)






4. Notch and S-N Line







×

Stress-Life Fatigue and Goodman Diagram

Method: Mean Stress Criteria, S-N Line, What It Assumes, Common Mistakes

What the tool does

It takes one repeating stress cycle and the strengths of the material and tells you whether the part should last indefinitely, by four textbook criteria, and whether it yields on the first cycle. If the cycle is too severe for infinite life, it estimates the number of cycles to failure from a straight S-N line on log axes.

The idea: a cycle is a point, a criterion is a line

A constant-amplitude cycle between a maximum stress σmax and a minimum σmin is described by two numbers: the alternating stress σa = (σmax − σmin) / 2, half the range, and the mean stress σm = (σmax + σmin) / 2. Two ratios say the same thing in another way: the stress ratio R = σmin / σmax (R = −1 fully reversed, R = 0 zero-to-tension) and the amplitude ratio A = σa / σm.

Plot the mean stress across and the alternating stress up, and the cycle is one point. With no mean stress the part survives indefinitely if σa is below the endurance limit Se. With no alternating stress it survives if σm is below the static strength. Each criterion joins those two ends with a line; a point inside the line is predicted to last.

Fatigue diagram for a steel with endurance limit 50 ksi, yield 80 ksi and ultimate 100 ksi: the modified Goodman, Gerber, ASME-elliptic, Soderberg and yield lines, with a cycle of 35 ksi alternating and 40 ksi mean stress and its load line from the origin
Se = 50 ksi, Sy = 80 ksi, Sut = 100 ksi; cycle of 35 ksi alternating on 40 ksi mean. The load line meets the Goodman line (open circle) at 0.909 of the way to the point, so the Goodman factor of safety is 0.909; the Gerber line is met beyond the point (1.13).

The four criteria

Written as in NASA TM 103644 (equations 1 to 3), with Sut the ultimate and Sy the yield strength:

  • Modified Goodman: σa / Se + σm / Sut = 1. A straight line to the ultimate strength. The usual design choice; cautious for ductile metals.
  • Gerber: σa / Se + (σm / Sut)² = 1. A parabola to the ultimate strength. Runs through the middle of test data for ductile metals, so about half the tests fall inside it.
  • Soderberg: σa / Se + σm / Sy = 1. A straight line to the yield strength. The most cautious; it never allows yielding.
  • ASME-elliptic: (σa / Se)² + (σm / Sy)² = 1. A quarter ellipse to the yield strength.

Factor of safety. If the alternating and mean stress grow together (a load line through the origin), the point reaches a line when both are multiplied by n. Putting nσa and nσm into each equation and solving for n gives:

Goodman: n = 1 / (σa / Se + σm / Sut)
Soderberg: n = 1 / (σa / Se + σm / Sy)
Gerber: n = ½ (Sut / σm)² (σa / Se) [√(1 + (2 σm Se / (Sut σa))²) − 1]
ASME-elliptic: n = 1 / √((σa / Se)² + (σm / Sy)²)

First-cycle yield (Langer line). None of the lines to Sut knows about yielding. The maximum stress σa + σm must also stay below the yield strength: ny = Sy / (σa + σm). With a high mean stress and a small ripple this check, not fatigue, decides.

Compressive mean stress. The tool uses the horizontal line σa = Se for every criterion (n = Se / σa) and checks yield against the largest compressive stress, ny = Sy / (σa − σm). The Goodman and Soderberg lines continued to the left would give credit for compression; the Gerber parabola and the ellipse, being symmetric, would penalize it, which tests do not support (NASA TN D-3883 shows the Gerber parabola failing at negative mean stress).

The endurance limit

Se′ is the fully reversed stress a small polished specimen survives indefinitely. Where no test value is known, NASA TM X-73307 notes that one-half of the ultimate strength is often used. That rule is for steel; the cap of 100 ksi (700 MPa) for very strong steels is a textbook rule the tool states from memory. Aluminum has no true endurance limit: enter its fatigue strength at a stated long life as a tested value.

In Advanced mode the specimen value is lowered to the part by five factors, Se = ka kb kc kd ke Se′, for surface finish, size, load type, temperature and reliability. The tool holds no factor tables: each is a number you enter from your own handbook, and 1 means no change.

A notch raises the local stress. The tool multiplies the entered nominal stresses by the fatigue notch factor Kf, which you can get from the Notch Sensitivity tool. Applying Kf to the mean stress as well is the cautious default; the alternative applies it to the alternating stress only.

Finite life: equivalent stress and the S-N line

A cycle outside the Goodman line still has a life. First replace it by the fully reversed cycle that does the same damage. Putting Sr = σar in the equations of NASA TM 103644:

Goodman: σar = σa / (1 − σm / Sut)
Morrow: σar = σa / (1 − σm / σf′), with σf′ the fatigue strength coefficient
Walker: σar = σmax1−γ σaγ = σmax ((1 − R) / 2)γ; γ = 0.5 is the Smith-Watson-Topper stress √(σmax σa)

The Walker form is the equivalent stress of MIL-HDBK-5J, Seq = Smax (1 − R)A3 (equation 9.6.1.4(a)), divided by 2γ so that it equals the amplitude for a fully reversed cycle. The handbook fits the exponent to test data for each alloy, usually between 0.3 and 0.7.

Then read the life from the S-N line. On log axes the line is straight, S = a Nb, from f Sut at 10³ cycles to Se at the knee Nk (10⁶ cycles for steel):

b = log(Se / (f Sut)) / log(Nk / 10³), a = f Sut / (10³)b, N = (σar / a)1/b

With the knee at 10⁶ cycles this is a = (f Sut)² / Se and b = −(1/3) log(f Sut / Se). The fraction f = 0.9 is a textbook estimate stated from memory; it is lower for high-strength steel.

What it assumes

  • One constant-amplitude cycle of uniaxial stress. For a load history with many levels use Spectrum Fatigue Life, which works from the MIL-HDBK-5J test curves.
  • The stress stays elastic and the life is long (above about 10³ cycles). Below that use Low-Cycle Fatigue.
  • Alternating and mean stress grow in proportion when the load grows. If the mean stress is a fixed preload and only the alternating part grows, the factor of safety is different (read it vertically on the diagram).
  • The same yield strength in tension and compression, and mean (50 % survival) strengths unless you enter a reliability factor.
  • For a rotating shaft with bending and torsion together use Shaft Design, which combines the stresses first.

When it misleads

  • Scatter. Fatigue lives of identical parts differ by a factor of several. The life printed is a mean; design and certification divide it by a scatter factor or use reduced curves.
  • Estimated endurance limit. One-half of ultimate is a rule for steel in rotating bending. Welds, castings, plated or corroded surfaces and non-ferrous metals can be far below it.
  • Steep sensitivity. With b near −0.09 a 10 % error in stress changes the life by a factor of about 3. Treat the cycle count as an order of magnitude.
  • Notches and yielding. Kf on the mean stress overstates it once the notch yields; ignoring it understates it if the notch stays elastic.
  • Gerber and the ellipse are fits through the middle of data, not lower bounds. A factor of safety of 1.0 by Gerber means even odds.

Common mistakes

  • Using the stress range as the alternating stress. σa is half the range.
  • Comparing the maximum stress with the endurance limit. Se is an alternating stress at zero mean; it is compared with σa, after the mean-stress correction.
  • Forgetting the yield check. A point can sit inside the Goodman line and still exceed the yield strength on the first cycle.
  • Using the polished-specimen endurance limit for a machined, notched or large part without the factors.
  • Applying the theoretical factor Kt where the fatigue factor Kf belongs, or applying a notch factor to Se and to the stress as well (once is enough).
  • Extending the S-N line below 10³ cycles or reading a life for a point inside the endurance limit.
  • Mixing R and A, or taking R = 0 to mean fully reversed. Fully reversed is R = −1.
  • Treating a compressive mean stress as harmful because the Gerber or elliptic formula, taken literally, says so.



References

Kalluri, S., and McGaw, M. A. (1991). Effect of Tensile Mean Stress on Fatigue Behavior of Single-Crystal and Directionally Solidified Superalloys. NASA Technical Memorandum 103644, Lewis Research Center, Cleveland, OH.

Figge, I. E. (1967). An Empirical Equation Relating Fatigue Limit and Mean Stress. NASA Technical Note D-3883, Langley Research Center, Hampton, VA.

National Aeronautics and Space Administration (1975). Astronautic Structures Manual, Volume III. NASA Technical Memorandum X-73307, Marshall Space Flight Center, AL.

U.S. Department of Defense (2003). Metallic Materials and Elements for Aerospace Vehicle Structures. MIL-HDBK-5J, Department of Defense, Washington, DC.

Budynas, R. G., and Nisbett, J. K. (2015). Shigley's Mechanical Engineering Design, 10th ed. McGraw-Hill Education, New York.


DISCLAIMER: The methods and calculation on aNOWysis.com are provided as means to educate the reader. There is no warranty, expessed or implied, as to the relevance or accuracy of the methods or calculations. All calculations should be checked by a qualified analyst for appropriateness and accuracy. We will not be held liable for any damages, injuries or deaths caused by using methods or calculations from this site.