Tools › Structures › Damage Tolerance › Stress Intensity Factors › Method
Stress Intensity Factors: Method
Open the Stress Intensity Factors tool
Stress Intensity Factors: the Newman–Raju Equations, the Through-Crack Forms, Superposition and Compounding
What the tool does
The stress intensity factor K measures the severity of the stress field at a crack tip. Every fracture and crack-growth calculation starts from it: fracture when K reaches the toughness, growth per cycle as a function of the range of K. The tool evaluates K for fourteen crack cases, prints the geometry factor behind each number, and with an entered toughness gives the margin at limit load, the stress at fracture and the critical crack size. It also tabulates the geometry factor against crack size and fits a cubic that the Crack Growth tool accepts.
Conventions
- K = β S √(π a): S is a remote stress on the gross section and a the crack size. β is the geometry factor; it is 1 for a through crack of half-length a in an infinite plate under tension.
- Part-through cracks have an elliptical front. a is the size in the thickness direction and c the size in the width direction. The parametric angle φ of the ellipse is 0 at the c-tip and 90° at the a-tip. For a surface crack the a-tip is the deepest point and the c-tip is where the front meets the surface; for a corner crack the a-tip is on the edge face (or the hole bore) and the c-tip on the plate face; for an embedded crack and a crack on a hole bore the crack is centered in the thickness.
- t is the full plate thickness and W the full plate width; 0 for W means an infinite width. Cracks and holes are on the centerline of the width; the corner crack is at one edge. (The report writes its equations with t as half the thickness for the embedded crack and the crack on a bore, and with b as half the width for every case but the corner crack; the tool converts.)
- Bending stress is the outer-fiber stress, through the thickness for part-through cracks (crack on the tension face) and in the plane of the plate for the single edge crack. Bearing stress is the pin load over hole diameter times thickness.
- Stress intensities are plain numbers in ksi√in or MPa√m, chosen by a select that does not follow the unit switch (1 ksi√in = 1.0988 MPa√m).
- Load level: a cracked structure is required to carry limit load, so the fracture margin is Kc / K at limit load − 1 and its box says LIM, as in the Residual Strength tool. Entered ultimate loads are divided by the factor of safety first.
Part-through cracks: the Newman–Raju equations
Newman and Raju fitted equations to three-dimensional finite-element results for five configurations (NASA TM 85793). All have the form
K = (St + H Sb) √(π a / Q) F, F = (M1 + M2 (a/t)² + M3 (a/t)⁴) g fφ fw, H = H1 + (H2 − H1) sinp φ.
Q = 1 + 1.464 (a/c)1.65 for a/c ≤ 1 (c/a in place of a/c beyond) approximates the square of the complete elliptic integral of the second kind. fφ is the angular function of the elliptical crack in an infinite solid, fw a finite-width correction, the bracket the correction at the a-tip, and g one or more fine-tuning functions. For cracks at a hole one of the g functions carries the stress concentration of the hole, through λ = 1 / (1 + (c/R) cos(μ φ)). H1 and H2 are the bending multipliers at φ = 0 and 90°.
The tool reports β = F / √Q for tension and H F / √Q for bending, so that every case reads K = β S √(π a).
- Surface crack (eqs 14 to 36): tension and bending, a/c up to 2.
- Embedded elliptical crack (eqs 3 to 13): tension only; the report developed no bending equation.
- Corner crack (eqs 37 to 52): tension and bending, a/c from 0.2 to 2.
- Surface crack on the bore of a hole (eqs 53 to 64): tension only.
- Corner crack at a hole (eqs 65 to 94): tension and bending.
The hole equations are for two symmetric cracks. One crack is estimated, as in the report, by Shah's factor √((4/π + a c / (2 t R)) / (4/π + a c / (t R))) on the two-crack value for an infinite plate, followed by the width correction for one crack.
Range enforced (Table 1 of the report): a/t below 1 (below 1.25 (a/c + 0.6) for surface and embedded cracks with a/c below 0.2); a/c from 0.2 to 2 for corner cracks and cracks at holes; R/t from 0.5 to 2; crack and hole within half of the half-width; a/t up to 0.8 for a corner crack at a hole in bending. The report states that the accuracy beyond a/t = 0.8 has not been established; the page says so when the crack is that deep.
A negative bending value means the bending stress closes that part of the front. The report allows its use only while the total of tension and bending stays positive.
Through cracks
- Center crack, half-length a, width W: β = √(sec(π a / W)), and the modified form β = (1 − 0.025 x² + 0.06 x⁴) √(sec(π x / 2)), x = 2a/W, which follows Isida's series solution more closely. Against the collocation values of NASA TN D-6376 the modified form is within 0.15 % up to 2a/W = 0.8 and the plain secant within 1 %.
- Single edge crack, length a, with θ = π a / (2W): tension β = sec θ √(tan θ / θ) (0.752 + 2.02 (a/W) + 0.37 (1 − sin θ)³); in-plane bending β = sec θ √(tan θ / θ) (0.923 + 0.199 (1 − sin θ)⁴). These are the forms printed in the USAF Damage Tolerant Design Handbook.
- Double edge crack, each of length a: β = (1.122 − 0.561 x − 0.205 x² + 0.471 x³ − 0.190 x⁴) / √(1 − x), x = 2a/W. It returns 1.122 for short cracks and the exact deep-crack limit (2/π) / √(1 − x).
- One crack at a hole, length a from the edge of a hole of diameter D on the centerline: with z = 1 / (1 + 2a/D), tension β = f0(z) Gw and pin bearing β = (f0(z) D / (2W) + f1(z)) Gw, Gw = √(sec((π/2)(D + a)/(W − a))), with the polynomials f0 and f1 of the USAF handbook. f0 runs from 3 × 1.12 at a = 0 to 1/√2 for a long crack.
- Two cracks at a hole: Newman's collocation values (NASA TN D-6376) with the width correction √(sec(π R / W) sec(π (R + a) / W)), which is the correction of the report for a/t = 1. It stays within 3 % of Newman's finite-plate values up to 2 (R + a) / W = 0.78 for holes up to half the width.
- Point loads on the crack faces: a pair of opposite forces P at x0 from the center of a crack in an infinite plate gives K = P / (t √(π a)) √((a + x0) / (a − x0)) at the nearer tip and the inverse root at the far tip. K falls as the crack grows.
Superposition
For one crack in one body, the stress intensities of separate loadings add. Tension and bending add in every case above. The pin-loaded hole is built this way: a pin load P reacted on one side equals half of (a remote stress P / (W t) on both sides) plus half of (a pair of opposite pin loads), which gives the bearing factor above; a bypass stress is added as a separate tension. The point-load solution is the kernel for any pressure σ(x) on the crack faces: K = (1 / √(π a)) ∫ σ(x) √((a + x) / (a − x)) dx, and since a crack in a stressed body has the same K as the crack loaded by the stresses the uncracked body carries on the crack line, this covers any stress field. Negative totals mean a closed crack, not a negative K.
Compounding
For boundaries that are not in a listed solution, the usual first estimate multiplies the factors of each boundary taken alone: β ≈ β1 β2 …, for example a hole factor times a width factor. It is an approximation, good when the boundaries are well apart and poor when they interact strongly; the two-crack width correction above is such a product, and its comparison with a full solution shows the size of the error. A case that matters should be checked against a boundary-element or finite-element solution.
Fracture check, critical size and the table
With an entered toughness the margin is Kc / Klim − 1, Klim being the largest K on the crack at limit load. Because K is linear in the loads, the loads at fracture are the limit loads times Kc / Klim, all raised in proportion. The critical crack size is found by a root search on K = Kc at limit load along the growth path: a through crack lengthens; a part-through crack is taken to keep its aspect ratio a/c. The table gives β against a on the same path and a least-squares cubic in a / W (a / t for part-through cracks), which is the form of the entered-β geometry of the Crack Growth tool.
What it assumes
- Linear elastic material and a crack-tip plastic zone small against the crack and the ligament. The net section is not checked for yielding here; the Residual Strength tool does that.
- Mode I only: the crack plane is normal to the stress.
- Flat plates of uniform thickness, uniform remote tension or linear bending, cracks and holes on the centerline, open holes (no interference, no cold work, no fastener clamp-up). The bearing case is a pin load on a through crack only.
- An elliptical crack front, and for the critical size and the table a constant aspect ratio. Real fatigue cracks change shape as they grow: a/c follows the ratio of K at the two tips. Growth of both axes is integrated in the Crack Growth tool.
- Poisson's ratio 0.3 in the finite-element results behind the fits.
When it misleads
- The equations are fits. Against the finite-element tables printed in the NASA reports they are within about 7 % for the surface, embedded and corner cracks in tension, 8 % for the surface crack in bending, 11 % for the crack on a hole bore and 14 % for corner cracks at a hole (the equation above the 1978 values where the front meets the plate face). They are least accurate for deep cracks (a/t of 0.8 and beyond) and where the front meets a free surface.
- Corner crack in bending: for a/c of 0.2 at a/t of 0.8 the equation is 18 % above the finite-element value at the plate face. For a/c of 2 the values tabulated in the report itself (its Table 2) are 10 to 31 % above its equation, and above its own tension values at a/t = 0.2, which bending cannot be; the tool follows the equation as the report plots it, and that case should be treated with caution.
- Later finite-element work (the Raju–Newman 1988 corner-crack values printed in the USAF handbook) is up to 27 % below the 1984 equation for a long shallow corner crack at a/t = 0.8. The 1984 equations are on the high side there.
- One crack at a hole by Shah's factor is an estimate, and the single through crack at a hole assumes the hole on the centerline.
- Toughness depends on thickness, orientation and temperature. A part-through crack in a thin sheet is not in plane strain, and KIc is then a conservative choice; a through crack in sheet needs Kc at that thickness or an R-curve.
- A part-through crack that reaches the back surface becomes a through crack with a different K; the solutions stop before that.
- Fastener holes in service carry clamp-up, interference and residual stress, all of which change K at short cracks.