A B S T R A C T The well-known trade-off between strength and fracture toughness in bulk specimens is often used to explain the low fracture toughness of very thin ductile face-centred cubic metal specimens, but this interpretation contradicts the relative length scales of thickness-dependent strength and thickness-dependent fracture toughness. This study uses the concept of similitude to demonstrate that linear elastic fracture mechanics analysis of 25.4 μm thick annealed aluminium is invalid, although the resulting fracture toughness measurements fit well with the existing literature and idea of a strength/fracture toughness trade-off. Similarly, an elastic plastic fracture mechanics analysis is sensitive to out-of-plane deformation that cannot be practically eliminated or corrected for with a model. However, a plastic collapse analysis using a critical net section stress criterion is demonstrably valid by the concept of similitude, is insensitive to out of plane deformation, and agrees with the evidence of extensive plasticity in the fracture surfaces.Keywords ductile tearing; fracture mechanics; fracture toughness; plastic collapse; stable crack growth; thin sheet. N O M E N C L A T U R Ea = crack length in SE(T) specimen or half crack length in M(T) specimen Atot = total work applied to specimen A tot ¼ ∫ Δ 0 Pd Δ b = uncracked ligament length B, B 0 , B * = thickness, minimum thickness for plane strain, critical thickness J, J max = J-integral, maximum valid J measurement for a given specimen goemetry K = stress intensity factor K Ic , K R = fracture toughness in mode I loading, propagation toughness in K-R analysis L À = mean lineal intercept grain size LEFM, EPFM = linear elastic fracture mechanics, elastic-plastic fracture mechanics M(T) = middle crack tension specimen geometry P = force applied to specimen R = load ratio σ min /σ max used in fatigue precracking SE(T) = single edge crack tension specimen geometry t = traction acting normal to J-integral path Γ u = displacement vector at a point along path Γ w = width of SE(T) or half width of M(T) specimens W = strain energy density x, y = in-sheet coordinate axes perpendicular and parallel to tensile stress, respectively Y = geometric correction factor used in computation of K ¼ σ Y ffiffiffiffiffiffi π a p Correspondence: C. L. Muhlstein.

A B S T R A C T In this work we present a strain field mining methodology where the intensities and spatial distributions of strains in double lap strap epoxy (EA 9394) bonded joints are measured, analysed and correlated with bond strength. While the global behaviour of the bonded joints was linear elastic at room temperature, discrete regions of elevated (relative to the full field average) shear strains were observed. The size, spatial distribution and other features of these strain 'hot spots' were described with a metric called the homogeneity index. This index captures the uniformity of the strain field with respect to regions of positive and negative strain at a specific stress level. The linearity of the progression of the homogeneity index and its magnitude was correlated with a joint's fracture strength and failure mode. The index was a robust predictor of double-lap strap bonded joint performance between 30% and 90% of a specimen's strength and can be used to improve joint design, manufacturing, quality assurance, maintenance and inspection.A = pixel area of a hot spot A1 = fitting parameter A2 = fitting parameter B1 = fitting parameter B2 = fitting parameter C1 = fitting parameter C2 = fitting parameter τ ult = average failure shear stress of the adhesive for all specimens δ centre = nearest neighbour distance based on the centroids of hot spots δ edge = nearest neighbour distance based on the edges of hot spots δ min = smallest resolvable displacement for digital image correlation ε min = minimum detectable strain between two features for digital image correlation ε th = threshold strain used to define a hot spot σ ia x = shear stress at the adhesive and inner adherend interface σ oa x = shear stress at the adhesive and outer adherend interface τ ult = failure shear stress of the adhesivẽ δ þ À centre = median nearest neighbour distance based on the centroids of hot spots δ þ À edge = median nearest neighbour distance based on the edges of hot spots à = median pixel area of a hot spot A coh = percent of epoxy failure ANOVA = analysis of variance A ROI = pixel area of the field of view σ x = average bond line shear stress DICT = digital image correlation and tracking Correspondence: C. L. Muhlstein.

No abstract

No abstract

No abstract