Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

www.springerlink.com/content/1738-494x

DOI 10.1007/s12206-011-0826-9

A short summary on finite element modelling of fatigue crack closure

Konjengbam Darunkumar Singh1,*, Matthew Roger Parry2 and Ian Sinclair3
Department of Civil Engineering, Indian Institute of Technology Guwahati, India 2 Landing Gear Integration, Airbus Operations Ltd, Bristol, U.K. 3 Materials Research Group, School of Engineering Sciences, University of Southampton, Southampton, U.K. (Manuscript Received February 3, 2011; Revised April 4, 2011; Accepted August 17, 2011) ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------1

Abstract
This paper presents a short summary pertaining to the finite element modelling of fatigue crack closure. Several key issues related to finite element modelling of fatigue crack closure are highlighted: element type, mesh refinement, stabilization of crack closure, crack-tip node release scheme, constitutive model, specimen geometry, stress-states (i.e., plane stress, plane strain), crack closure monitoring. Reviews are presented for both straight and deflected cracks.
Keywords: Fatigue crack growth; Variable amplitude; Crack closure; Finite element modelling. ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1. Introduction
Initial attempts in finite element modelling of fatigue crack growth were reported by Refs. [1, 2]. Comparison was made of cyclic and monotonic loading with a particular focus on changes in strain and stress distributions, and crack opening displacements upon release of the crack tip node (i.e., extension of the crack). Premature contact of the crack surfaces (i.e. crack closure) was observed during unloading in a tensiontension cycle. Ref. [3] studied closing and opening behavior, as well as stress and strain around the crack tip, for a growing crack using kinematic hardening theory under plane stress conditions. The crack was made to extend incrementally at every loading cycle, starting from a given crack length. Under a stepped program load, they observed a very high value of the effective stress intensity factor range just after the stress amplitude is raised, followed by a gradual decrease to a certain stable value, agreeing with the experimental results of Ref. [4]. Ref. [5] modified an existing finite element code to account for changing boundary conditions, crack growth and intermittent contact of crack faces under constant amplitude and simple block loading. Although a simple FE model was adopted, the calculated crack opening stresses under either constant or simple block program loading were found to be qualitatively comparable with experimental observations. It was observed that for higher values of load ratio, R (~0.45)
This paper was recommended for publication in revised form by Associate Editor Vikas Tomar * Corresponding author. Tel.: +91 361 258 2423, Fax.: +91 361 258 2440 E-mail address: darun@iitg.ernet.in KSME & Springer 2011

under constant amplitude loading crack opening levels were lower than the minimum applied stress levels, consistent with the experimental results by Ref. [6]. Calculated crack growth using Elbers crack growth equation (da/dN = C(Keff)m, where da/dN is the crack extension per cycle; Keff = Kmax Kcl, Kmax is the maximum stress intensity factor, Kcl is the stress intensity factor at which contact of crack faces occurs) from FE based opening levels gave qualitatively consistent agreement with experimental observations [7], i.e., crack growth retardation following a high block loading and acceleration following a low block loading. Ref. [8] analyzed stress and strain fields of a propagating fatigue crack and the resulting crack opening and closing behavior, finding that the strain range value in the vicinity of the crack tip is closely related to the effective stress intensity range determined on the basis of the modelled crack opening and closing behavior at the tip. Numerous other finite element studies of the crack closure phenomenon under different loading conditions, stress states and geometries have in fact been reported in the literature, including Refs. [9-44]. As may be found from the finite element studies in the literature, that for a consistent and reliable result, careful attention needs to be given to a series of decisions on mesh refinement, crack advance technique and attainment of a steady-state crack closure level. Further, it has been cautioned by Ref. [45] that it may not be best to place too much emphasis on the exact computed numerical values of closure or opening levels; rather, the focus should be to study the effects of various physical parameters on crack closure levels upon the attainment of a stable numerical model. In the following sections, these issues affecting FE modelling are briefly described.

3016

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

1.1 Element type and mesh refinement It is generally considered that higher order elements perform better in capturing the stress and strain fields near the crack tip. However, due to the computational cost demanded by higher order elements (from higher bandwidth requirement), non-linearity analysis needing many load steps and possibly easier meshing, early (e.g. Refs. [1, 3, 5, 8-12, 46, 47]) and some recent investigators (e.g. Refs. [22, 38, 44]) adopted constant-strain triangle (CST) elements to study fatigue crack propagation. It has been put forward that their results might not give an accurate result due to lack of mesh refinement and potential plane strain locking [49] of the elements [39]. When locking occurs, the stresses oscillate wildly from one element to another. To meet the incompressibility requirement associated with large plastic strains with such CST elements, Ref. [20] adopted a union-jack mesh configuration (i.e., the CST elements are arranged to form a series of squares and their diagonals) as suggested by Ref. [49]. Later workers (e.g. Refs. [23, 26, 31, 37- 39, 44, 50-55]) used predominantly 4-noded (Q4) linear strain elements. Ref. [23] found that Q4 elements predicted comparatively higher opening levels, apparently due to a better ability to capture the crack singularity leading to larger residual deformations. Recently, Ref. [39] reported a mesh convergence study of various elements, i.e., CST, CST with union jack configuration, Q4 and Q4 with the reduced integration method (considered to be helpful for avoiding plane strain locking [49]) for both the CCT (center crack tension) and CT (compact tension) test specimen geometries. It was observed that for the CCT geometry under plane strain condition, the steady-state opening stresses (steady-state opening stress values were obtained after the crack had grown approximately twice the initial forward plastic zone size) converged as the mesh was refined. However, for the CT geometry, the opening values did not converge (opening values were seen to decrease) with mesh refinement, indicating that PICC is negligible or does not exist under plane strain conditions, whereas for plane stress, convergence in opening values was seen readily for both CCT and CT geometries with Q4 elements. Refs. [52, 53, 56-59] adopted Q4 elements, which use the selectively reducedintegration technique (whereby reduced integration was used for volumetric strain, and full integration for deviatoric strain). Parry specifically verified that these elements could model the incompressible material behavior under plane strain. Attempts have also been reported using higher order elements namely, triangular elements of cubic base functions (e.g. Ref. [18]) and Q8 (8-noded quadrilateral) elements (e.g. Refs. [43, 51, 54, 61]). Ref. [51] conducted a comparison of Q4 and Q8 (8noded quadrilateral) elements. A sinusoidal residual stress pattern, with corner nodes in compression and mid-side nodes in tension was observed in the crack wake when Q8 elements were used. Refining the mesh reduced the amplitude of sinusoidal pattern, but failed to eliminate the compression-tensioncompression residual stress pattern along the edge of the ele-

ment. The behavior is attributable due to the displacement functions used for Q8 elements, resulting in non-uniform stiffness along the element edge. However, for Q4 elements, mesh refinement, reduced the amplitude of the saw-tooth pattern of stresses to an acceptably low magnitude. Meshing along the crack line must be sufficiently refined to capture the severe stress concentration at the crack tip and to simulate real crack growth. However, extreme mesh refinement would lead to a large stiffness matrix bandwidth and many degrees of freedom demanding long run times. Furthermore, many increments of crack growth would be required for crack propagation studies. Optimizing with these two constraints is required for a computational cost effective and reliable result. Ref. [9] studied three meshes composed of constant strain triangles of different sizes on the crack line and identified critical mesh spacings for dependable solutions at different stresses. Ref. [23] observed opening levels for different meshes for a wide range of stress levels and two stress ratios. They found that above certain stress levels at each Rratio, the meshes considered ( Le W = 0.002 and Le W = 0.004 , where W is the half plate width, and Le is the element size along the crack line) gave reasonably similar results, with a consistent tendency for the finer mesh to yield higher opening levels. At intermediate maximum stress values, normalized opening levels (Sopen/Smax) sharply increased and then sharply decreased with Smax/o, where S denotes remotely applied stress and o is the yield stress. They suggested this to be a false peak, i.e., an artefact of the modelling process, noting that at low loads the mesh is not sufficiently fine to capture reversed yielding upon unloading. They suggested a simple criterion for mesh sufficiency based on the ratio of the element size along the crack line to the forward plastic zone size ( rp ). They recommended that a value of Le rp 0.05 is good enough for reliable results with Newmans constant strain elements, while for first order quadrilateral elements (Q4), a criterion of Le r p 0.10 seems to be sufficient for R = 0 , and Le rp 0.15 for R = 1 . In a later study Ref. [26] observed the criterion to be slightly overly conservative for plane strain where plastic zones widths along the crack propagation line are smaller and the near-tip has severe strain gradients, and suggested Le rp 0.20 be satisfied. However, mesh refinement studies on CCT and CT geometries by Ref. [51] under plain strain suggested that the mesh density ahead of the crack should satisfy Le r p 0.10 . Ref. [61] suggested from their FE study on the CCT geometry with Q4 elements that, a unique, most-appropriate mesh size exists for a given loading condition that will give good numerical results with close accord with the experimental results. The value of the most-appropriate mesh size was found to be nearly constant, ranging from 0.77 to 0.91 times the reversed plastic zone size (rc); however, in terms of the monotonic plastic zone size it was seen to decrease with increasing Kmax, for the ranges of R (0-0.3) considered, suggesting that rc may relate better for estimating appropriate mesh size. However, Ref. [39] argued that this method is flawed as various techniques [e.g. Refs.

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

3017

[62-66]] give inconsistent results in measuring the opening loads. Ref. [39] suggested that an adequate mesh (Q4 elements) refinement is one which contains 3-4 elements in the reversed plastic zone. However, a slightly coarser element of about 0.5-1.0 rc was suggested by Ref. [43] using Q8 elements. Ref. [43] found that elements that are too large give reduced closure levels because of the inability of the elements to capture the crack tip elasto-plastic behavior, while very small size elements also tend to give lower closure levels. A recent 3D small scale yielding fatigue crack growth study reported that computed opening loads showed an independence of mesh refinement when a) the plastic zone at peak load encloses more than 10 eight-noded brick elements, b) at least two elements are contained in the reversed plastic zone, and c) the half-thickness has at least five element layers [41]. Thus, from the findings of Refs. [39, 41], about 20 elements in the forward plastic zone are suggested for simulating fatigue crack closure using the FE method. However, studies by Ref. [52] observed that mesh refinement of Le = rp/30 resulted in slightly lower crack opening profiles in the region away from the near-tip, and higher opening profiles in the near-tip region as compared to that of Le = rp/15 refinement, suggesting that mesh refinement may have an influence in the closure determination. Nevertheless, when exact quantitative corroboration between experimental and FE results is inherently difficult considering the idealized nature of FE models, and for long fatigue crack growth simulation wherein the computational cost is expensive, a mesh refinement of a minimum of about Le = rp/10 (as adopted by many researchers e.g., Refs. [22, 37, 44, 51, 54]) is expected to give fairly good understanding of the closure phenomena. While most the aforementioned criteria on mesh refinement requirement are expressed in terms of the element length along the crack plane (Le), Ref. [51] found that an aspect ratio (He/Le, where He is element height) of 2 significantly reduced the amplitude of saw tooth residual stress pattern along the crack wake, and this consideration is adopted by Ref. [37]. Although they attributed the benefit of using an aspect ratio of 2 to the bending resistance of the crack plane element immediately ahead of the tip, it may be mentioned that elements He/Le 1.0 (i.e., square element) are considered a better choice and help in minimizing interpolation errors (Ref. [43]). 1.2 Stabilization of crack closure In many cases, cracks progressing from an initial length are reported to reach a constant closure level (e.g. Refs. [23, 44]). To attain this stabilization, the crack must grow far enough to develop a significant plastic wake and the crack tip should be outside the stress field of the initial defect or pre-crack. Refs. [43, 61] observed stabilized closure levels after a crack has propagated ahead of the monotonic plastic zone of the first cycle. However, Refs. [20, 67] reported non-stabilized crack opening values even after the crack has propagated through the initial forward plastic zone. Ref. [68] observed that a

steady-state closure level will typically be obtained when a crack has propagated through three or four overload plastic zone sizes flowing a transient. Ref. [23] suggested that a modelled crack should be allowed to grow beyond the region of material which experiences significant plastic deformation on the first cycle of loading. A large amount of crack growth and sufficiently high applied stress have been found to show an initial stabilization followed by a subsequent decrease in crack opening levels due to the loss of elastic constraints resulting from extensive yielding around the crack tip [69]. A similar observation on decaying opening levels when the remaining ligament becomes small has also been reported [70]. 1.3 Crack-tip node release scheme There is no clear consensus on the appropriate FE mechanisms for crack advance, with various schemes being reported in the literature. Again, none of the reported approaches can be assumed to represent the fatigue process where, according to slip models of striation formation, crack propagation is a progressive process occurring during the entire loading cycle (Refs. [33, 43]). Attempts have been made to develop crack advance schemes using critical damage criteria, such as Refs. [1, 2] critical cumulative plastic work near the tip, Ref. [10] critical strain criterion, Ref. [13] critical crack opening stress, and the critical energy rate criterion of Ref. [71]. Other reported crack propagation schemes include moving mesh methods (e.g. Ref. [14]), cohesive element methods (crack propagation is regarded as a gradual process in which separation between incipient material surfaces is resisted by cohesive forces) (Ref. [71, 72]) and X-FEM (extended FE method, where discontinuous and 2D asymptotic crack tip displacement fields are added to the FE approximation to account for the crack, thus enabling the domain to be modelled with no explicit crack surface meshing) [74]. However, most numerical models adopt the technique of releasing a model node (e.g., extension of one element) on every cycle for efficiency (e.g., Refs. [44, 45]) and few have tried releasing more than one node per cycle (e.g., Refs. [43, 44, 60]). Ref. [43] found that releasing nodes at every cycle resulted in smoother crack profiles. Refs. [5, 9, 10, 16, 17, 19, 32, 33, 39, 43, 52, 55, 71, 75-77] have adopted the release of the crack tip node at maximum load in each cycle to simulate crack growth. Refs. [3, 8, 11, 12, 15, 26, 39, 61, 74, 79] alternatively released the crack tip node at the minimum load in each fatigue cycle. Ref. [80] contended the rationality behind propagating a crack at a minimum load cycle, as the crack tip would obviously be subjected to compressive stress field. Refs. [113, 14] use a scheme of releasing the crack tip node at different points along the forward loading excursion, and observing from their formulation that the apparent crack closing levels were dependent on the timing of the node release. Refs. [81, 82] suggested a criterion of extending the crack at 98% of the maximum load. Ref. [59] suggested release of node after the applied load has decreased slightly from peak

3018

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

value for a stable iteration process. Refs. [19, 44, 50] considered a node release scheme in which the crack tip was allowed to propagate immediately after the point of maximum load, during the first increment of unloading. Ref. [22] employed the gradual release of the crack tip node during unloading, beginning at the maximum load. However, studies on threedimensional semi-circular fatigue crack by Ref. [32] suggested that node releases at 10%, 50% and 100% of the maximum load resulted in very little effect both on the surface and in the interior of the specimen. There are further reports of observed differences when using either the maximum or minimum load node release scheme (e.g. Refs. [23, 61, 67]). McClung and Sehitoglu [23] compared three schemes-node release at maximum load with stabilization (after Newman), node release at minimum load (after Ref. [11]), node release immediately after maximum loading (after Ref. [50]), and observed that the differences in the three schemes were not significant, especially at higher stresses. However, Ref. [67] found some variation in the crack opening values under the minimum and maximum loading node advance schemes. Ref. [61] observed that releasing the node at maximum load on every cycle produces higher opening levels as compared to closing levels (similar finding can be found in the work of Refs. [9, 33]), which is generally contradictory to experimental results; on the other hand the reverse is found to be true when nodes are released at the minimum load on every cycle. Ref. [40] suggested that the differences may be due to the insufficient discretization of the reversed plastic zone size. Ref. [78] revisited their earlier work [23] after truss elements (used to impose the changing boundary condition) were removed. When the boundary conditions on the crack line nodes were imposed directly, they observed that both the maximum and minimum node release schemes gave similar crack opening levels for nearly all the cases that were examined. Recently, Refs. [39, 40] showed that both the minimum and maximum load node release schemes agree well when a suitably refined mesh is used. Although it is now generally evident that node releases at maximum or minimum do not substantially affect numerical results concerned, releasing at maximum load cycle may appear to be closer to a realistic physical process, with crack propagation at minimum load generally carried out to avoid numerical solution difficulties [43]. 1.4 Effect of constitutive model The majority of finite element studies of crack closure in the literature have employed bilinear hardening stress-strain relationships due to the simplicity in numerical implementation (e.g. Refs. [20, 31, 37, 44, 78]). However, such relationships do not accurately represent the constitutive response of many real materials (although, some materials like medium carbon steels, may be reasonably well represented in response to cyclic loading [83]). Ref. [24] made an attempt to accurately simulate rounding of the stress strain curve on each load re-

versal following significant plastic deformation, based on the Ramberg-Osgood power-law formulation. Although this model is more sophisticated, it has its own limitations. The implementation in numerical analysis is more involved than a bilinear model, and accurate calibration of the material behavior is required. The power-law may be able to represent well a cyclically stable stress-strain relationship, and is probably therefore a better descriptor of near-tip behavior. It is seen that under different loading conditions and for different materials, both forms of material model may compare well with experimental results. For instance, the closure levels obtained from the experiments of Ref. [84] taken at relatively low maximum stress levels agree well with bilinear model results, whereas power-law hardening model results compare well with the experimental results of Ref. [85] for higher maximum stress levels. Refs. [18, 51] have used multi-linear stress-strain curves by explicitly defining the dependence of yield stress on increasing plastic strain. However, there appears to be a lack of detailed comparison between closure levels obtained using this approach and using a bilinear model. Ref. [43] studied isotropic and kinematic models (for plane stress conditions) and found that isotropic hardening predicted higher closure levels as compared to that of kinematic hardening models, attributing more significant plastic deformation with isotropic hardening model; however, the predicted closure levels using kinematic hardening models were found to be closer to experimental results. Ref. [52] reported for higher strength Alalloy modelling in SSY that there was no significant difference in crack opening displacements using kinematic, isotropic and elastic-perfectly plastic material behavior, suggesting that strain hardening effects and hardening may not be a primary factor in crack closure, at least for materials of interest here. 1.5 Effect of specimen geometry The influence of fatigue specimen geometry upon closure response was studied by Refs. [17, 26], conducting plain strain analyses for CCT and single edge notched bend (SENB) specimens. At a load ratio of 0, crack closure was observed for center-cracked panels but not for bend specimens. Closure was not observed for CCT specimens for R 0.3 . For a transient period of crack growth, as the crack evolved from the state of a stationary crack to the steady state of a growing fatigue crack, closure was enhanced for the CCT specimen format. They explained the influence of crack specimen geometry upon closure in terms of the T-stress. A decrease in closure levels has been particularly observed for increasing T-stress (Tmax/o is lower for CCT than SENB specimen) which would appeal, on first inspection at least, consistent with decreasing plastic zone dimensions as T rises. Ref. [30] also studied specimen geometry effects on fatigue closure. Elastic-plastic finite element analysis was conducted for CCT, single-edgecracked plate (tension), and single-edge-cracked plate (bend), at different crack length to width to ratios. It was observed that

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

3019

the maximum stress to flow stress ratio, Smax/o, which successfully describes closure in many CCT configurations, does not correlate with the different geometries. It was seen that the normalized stress intensity parameter, Kmax/Ko, where K o = o a , can correlate crack opening stresses for different geometries and crack lengths better. The quality of correlation was found to be good at small Kmax/Ko, and gradually degenerated as K max K o increase beyond the small-scale yielding regime. K of course becomes less effective in characterizing the crack tip deformation and stress fields at progressively higher stresses, consistent with a diminishing correlation with LEFM criteria (i.e., K based). 1.6 Plane strain and plane stress modelling of PICC It is generally agreed that extra material behind the crack tip may resist reverse deformation during unloading, causing a reduction in the crack growth rate. For plane stress conditions, a specimen may become thinner near the tip of a crack as material is transported to the crack flanks from the free surfaces of the sample. Each crack growth increment produces a part of the extra material which eventually extends over the entire crack length [87, 88]. Models by Ref. [89, 90] are able to describe this situation. As noted earlier, out-of-plane flow for plane strain conditions is not allowed, and there is no volume change of the material under plastic deformation by definition, hence various researchers are of the view that PICC cannot occur under these conditions [91, 92]. However, both experimental [93, 94] and numerical [11, 16] studies have been reported to suggest that PICC can occur under plane strain. Stabilized FE modelled closure levels were observed by Ref. [16], which were identified to agree well with results for experiments on Al 2024. However, mesh refinement criteria of Ref. [23] would not be met for the analysis performed by Ref. [11], hence the reliability of some results may be questionable. Investigations by Refs. [86, 88] suggest that plastic deformation left in the crack wake during fatigue crack propagation close the crack, near the tip, in spite of the volume constancy during plastic deformation, as a rotation of material occurs parallel to the growth direction. Two-dimensional finite element analysis reported by Ref. [17, 20] concluded, in general, that crack closure did not occur under steady-state plane strain conditions. Ref. [20] indicated that PICC in plane strain was due to an artificial wedge of material formed near the initial crack tip used in the modelling. It was put forward that PICC in plane strain exists only for a transient period due to the unrepresentative residual strains at the initial crack tip and its effect would diminish as the crack was allowed to propagate away from the wedge. This observation agrees with the findings of Ref. [57] where it has been observed that anomalous near-tip closure in FE models occurred under both plane stress and plane strain conditions, and is seen to vary with baseline load levels and crack propagation algorithms. In the case of plane strain models, propagation algorithms are seen to influence pre-crack closure. Further, it was concluded that none of the

plane strain models showed crack closure that could be related to ongoing/steady-state crack growth: crack closure in all cases was dominated by pre-crack contact, and/or anomalous near-tip contact, even for the relatively long crack propagation used in the low loading range models. These results are somewhat in contradiction to the later works of Refs. [26, 28, 33, 95], where closure levels on the order of 20% to 30% of Kmax are found, at a zero load ratio, under small scale yielding. Even at a higher load ratio of 0.5, closure has been reported in plane strain for 20% of the loading cycle [16, 18, 97]. Three-dimensional investigation reported by Ref. [21] has shown that a crack front may first close on a free surface (exterior) plane on the specimen, and close last on the interior plane, indicating that fully three-dimensional closure behavior corresponds to a continuous variation between twodimensional behavior for plane stress on the specimen surface, and plane strain in the mid-plane. 1.7 Monitoring crack closure Of the various ways of monitoring crack closure in FE models, the following may be identified as relatively simple and direct approaches: 1) observing the point in the unloading cycle at which the displacement of a crack wake node becomes zero (e.g. Refs. [5, 20]), 2) assessing the crack tip stress (e.g. Ref. [67]), 3) assessing the stress of the nodes behind the crack tip (e.g. Refs. [19, 23, 39, 44]) and 4) examining changes in specimen compliance (e.g. Refs. [15, 47]). Refs. [5, 20] have monitored the closure of the first node behind the crack tip; however it has been observed that, in plane strain, the first node behind the tip nearly may even close during the first unloading increment. As noted earlier, Ref. [23] suggested that it might be due to the incomplete redistribution of loads and plastic strains immediately upon node release. To circumvent this, they assessed closure by the behavior of the second node behind the crack tip, while opening levels were based on the behavior of the first node behind the crack tip. However, in a later paper [28, 44] returned to using closure of first node to examine closure. Ref. [39] assessed crack opening values based on the contact stress distribution at minimum load using superposition principle, and hence cannot be valid for large stress distribution. Ref. [67] assessed crack closure values based on the crack tip stress, employing the assumption that when the tensile stress borne by the crack tip node changed to compression, a crack can be considered close. However, this method depends on the accuracy of the crack tip stress where the stress gradients are at their most severe. Further, Ref. [77] argued that Ref. [67]s method may confuse the role of crack-tip residual stresses in fatigue crack growth (which are related to closure), as the presence of residual stresses at the tip does not imply crack closure. Ref. [43] observed that closure definition based on crack-tip stress reversal (from tension to compression) gives higher values of crack closure/opening stresses as compared to the conventional definitions based on first contact of the crack flank because a

3020

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

compressive stress state must exist for crack flank contact. Further, it was argued that contact of the first node behind the current tip does not indicate that the crack is fully open but that the crack is open between this node and the current crack tip. However, mesh convergence studies reported by Ref. [48] showed apparent merging of the results based on both definitions. Ref. [98] alternatively reported that definitions based on first node contact gave a better fit to the experimental results based on compliance change. Refs. [52, 53] determined closure points by neglecting the first node behind the crack tip and compared the results with closure points obtained using an offset compliance method, finding the results comparable, suggesting that either of the two methods could be used. Overall, significant ambiguity remains in monitoring the behavior of near-tip nodes and relating the results to real closure processes. 1.8 Numerical modelling of crack deflection Roughness-induced crack closure was first modelled numerically by Ref. [99] using finite difference (FD) techniques. It was assumed that the functional dependence of the displacements in time and space can be dissociated, giving rise to independent semi-discretizations for both time and space. The mesh of the continuum is made up of constant strain triangles and the computation is broken down into a number of time steps. Isotropic hardening and bilinear stress-strain curve were chosen. In the FD technique the continuum (elastic-plastic solid) is analyzed based on the conservation of momentum and the constitutive behavior of the material. The masses of the elements are assumed to be lumped at the nodes. In each time step, the forces acting at each node are used to calculate its acceleration, which is integrated to yield its velocity and displacements. These nodal displacements produce forces which result from: 1) relative displacements within each triangle giving rise to strains that cause the corresponding stresses through material properties, and 2) the contact between different surfaces. These forces, together with body forces and the external loads, give rise to the resultant force acting at each node. This completes one computational cycle and the process is continued until the final time step. Displacement boundary conditions are imposed by eliminating the equations of the nodes whose displacements are known. To effectively uncouple the equations of motion, the time increment between two time steps is made less than the time required for the information to travel between the two nearest nodes in a mesh (based on the pressure wave speed). To satisfy this criterion, numerical simulations were carried out at about 5000 Hz, which is comparatively large compared to the standard fatigue test frequencies (below 100Hz). Dynamic effects are claimed to be small and do not perturb the results. In Ref. [99]s model of RICC, it is stated that as the attention was focussed on modelling of RICC and the effects of plasticity were largely neglected, the source of crack closure was suggested to be in the displacements of the crack surfaces in the crack wake. Model-

ling criterion for PICC [23, 99], particularly the element size in the crack path being smaller than one-tenth of the plastic zone size, was not followed, and it was assumed that a ratio of crack branch length to element size of about four was reasonable. For the problem considering the maximum plastic zone size was about 1 m, giving the ratio of element size to plastic zone size of about 10:1 (which is much larger than suggested for PICC modelling, as discussed previously). It was observed that RICC closure was discontinuous and concentrated at points where the crack changes its direction of propagation. The main controlling factor of RICC was seen to be the deflection angle between the crack branches and the average direction of crack propagation. When the deflection angle was kept constant, crack closure took place in the last crack kink, and closure levels increased with the deflection angle (consistent with the geometrical model [100]), with no effect of the crack length for the range studied. However, when the deflection angle was allowed to change during crack propagation, significant amounts of closure developed, the crack being closed during the greater part of the fatigue cycle. The results were indentified to agree with high closure loads measured experimentally in 2124-T351 aluminium alloy [18]. The main mechanism of RICC was suggested to be when the mode I and mode II displacement ratio in the crack wake is different during loading and unloading steps. The proposed mechanism has been criticized on two points by Ref. [52] from a physical standpoint. First, the change in the mode I/ mode II displacements from loading to unloading during real crack propagation can be expected to be very small as, in a near threshold condition, the increment of crack growth becomes infinitesimal. Second, the mechanism would lead to residual deformation of a purely elastic crack, which cannot be realisticpermanent deformation of the crack faces and hence RICC can only occur from plastic deformation. Refs. [34-36, 52] investigated crack closure arising from crack deflection and plasticity using finite element modelling in a long fatigue crack in a notional aerospace aluminium alloy. The approach is an extension of Ref. [5] modelling concepts. The FE mesh consisting of four-noded quadrilateral elements and bilinear kinematic hardening model was used. The element length along the crack propagation direction was set so that Ref. [23]s mesh sufficiency criterion was met. An incremental crack propagation scheme was used along the lines of Refs. [5, 20]. Friction effects along the crack surface were not considered. Plane slip bands ahead of the crack tip were also simulated by restraining nodes in shear with rigidperfectly plastic truss elements. By controlling the yield stress of the truss elements, the effective critical resolved shear stress (CRSS) along the crack propagation direction was varied. Based on both overall specimen compliance and node behavior along the crack path, the results showed: 1) an increasing effect of crack deflection angle on RICC levels, consistent with the simple analytical model of Ref. [100] and FD model of Llorca [99], 2) strong dependence of closure on the residual plastic strains in the wake, rather than global shear displace-

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

3021

ments of the fracture surface due to mixed mode behaviour at the crack tip, and 3) low closure levels compared to experimental data, suggested to be consistent with the absence of environmental irreversibility in the FE models and the 2D idealisation of crack path. Slip band simulations showed a particular increase in closure levels at high deflection angles, improving the apparent accuracy of the modelling in relation to experimental data.

2. Conclusions
Several researchers have attempted to understand fatigue crack closure through finite element modelling approach. Although the FE approach is a convenient method, there are key issues one needs to address for a reliable result. A short summary of some of the important aspects of FE modelling of fatigue crack closure (both PICC and RICC) such as element type, mesh refinement, stabilisation of crack closure, crack-tip node release scheme, constitutive model, specimen geometry, stress-states (i.e., plane stress, plane strain), crack closure monitoring are presented in this paper.

Acknowledgment
This work was supported by Airbus, UK and EPSRC, UK. Helpful discussions with Drs. RA Collins, and J-C Ehrstrom from Alcan, France are gratefully acknowledged.

References
[1] H. Miyamoto, T. Miyoshi and S. Fukuda, An analysis of crack propagation in welded structures, Significance of Defects in Welded Structures, Proceedings of Japan-U.S. Seminar, Tokyo Press (1973) 189-202. [2] H. Miyamoto, The mechanics of crack propagation, Proceedings of symposium on mechanical behaviour of materials, Kyoto (1974) 37-47. [3] K. Ohji, K. Ogura and Y. Ohkubo, On the closure of fatigue cracks under cyclic tensile loading, International Journal of Fracture, 10 (1974) 123-124. [4] W. Elber, The significance of fatigue crack closure, In Damage Tolerance in Aircraft Structures, ASTM STP 486 (1971) 230-242. [5] J. C. Newman and H. Armen, Elastic-plastic analysis of a propagating crack under cyclic loading, AIAA Journal, 13 (1975) 1017-1023. [6] M. Katcher, Crack growth retardation under aircraft spectrum loads, Engineering Fracture Mechanics, 5 (1973) 793818. [7] W. T. Mathews, F. I. Baratta and G. W. Driscoll, Experimental observation of a stress intensity history effect upon fatigue crack growth rate, International Journal of Fracture, 7 (1971) 224-228. [8] K. Ohji, K. Ogura and Y. Ohkubo, Cyclic analysis of a propagating crack and its correlation with fatigue crack

growth, Engineering Fracture Mechanics, 7 (1975) 457-464. [9] J. C. Newman, A finite element analysis of fatigue crack closure, Mechanics of Crack Growth, ASTM STP 590 (1976) 281-301. [10] J. C. Newman, Finite element analysis of crack growth under monotonic and cyclic loading, Cyclic Stress-Strain and Plastic Deformation Aspects of Fatigue Crack Growth, ASTM STP 637 (1977) 56-80. [11] K. Ogura, K. Ohji and K. Honda, Influence of mechanical factors on the fatigue crack closure, Fracture 1977 (Editor: D.M.R. Taplin), 2 (1977a) 1035-1047. New York: Pergamon Press. [12] K. Ogura and K. Ohji, FEM analysis of crack closure and delay effect in fatigue crack growth in variable amplitude loading, Engineering Fracture Mechanics, 9 (1977b) 471480. [13] M. Nakagaki and S. N. Atluri, Fatigue crack closure and delay effects under mode I spectrum loading: an efficient elastic-plastic procedure, Fatigue Engineering Material Structure, 1 (1979) 421-429. [14] M. Nakagaki and S. N. Atluri, Elasto-plastic analysis of fatigue crack closure in modes I and II. AIAA 18 (1980) 1110-1117. [15] H. Nakamura, H. Kobayashi, S. Yanase and H. Nakazawa, Finite element analysis of fatigue closure in compact specimen, Mechanical Behaviour of Materials, Proceedings ICM4 2 (1983) 817-823. [16] A. F. Blom and D. K. Holm, An experimental and numerical study of crack closure, Engineering Fracture Mechanics 22 (1985) 997-1011. [17] N. A. Fleck, Finite element analysis of plasticity-induced crack closure under plane strain conditions, Engineering Fracture Mechanics, 25 (1986) 441-449. [18] R. O. Ritchie, W. Yu, A. F. Blom and D. K. Holm, An analysis of crack tip shielding in aluminium alloy 2124: a comparison of large, small, through-thickness and surface fatigue cracks, Fatigue and Fracture of Engineering Materials and Structures, 10 (1987) 343-363. [19] P. L. Lalor and H. Sehitoglo, Fatigue crack closure outside a small scale yielding regime, Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 342-360. [20] N. A. Fleck and J. C. Newman, Analysis of crack closure under plane strain conditions, Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 319-341. [21] R. G. Chermahini, K. N. Shivakumar and J. C. Newman, Three dimensional elastic-plastic finite element analysis of fatigue crack growth closure, Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 398-413. [22] T. Nicholas, A. Palazotto and E. Bednarz, An analytical investigation of plasticity induced closure involving short cracks, Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 361-379. [23] R. C. McClung and H. Sehitoglu, On the finite element analysis of fatigue crack closure 1, Basic modelling issues, Engineering Fracture Mechanics, 33 (2) (1989) 237-252.

3022

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

[24] R. C. McClung and H. Sehitoglu, On the finite element analysis of fatigue crack closure 2, Numerical results, Engineering Fracture Mechanics, 33 (2) (1989a) 253-272. [25] J. Llorca and V. Sanchez-Galvez, Modelling plasticityinduced fatigue crack closure, Engineering Fracture Mechanics, 37 (1990) 185-196. [26] R. C. McClung, Crack closure and plastic zone sizes in fatigue, Fatigue and Fracture of Engineering Materials and Structures, 14 (1991) 455-468. [27] R. C. McClung and D. L. Davidson, High resolution numerical and experimental studies of fatigue cracks, Engineering Fracture Mechanics, 39 (1) (1991) 113-130. [28] H. Sehitoglu and W. Sun, Modelling of plane strain fatigue crack closure. Transactions of ASME, Journal of Engineering Materials and Technology, 113 (1991) 31-40. [29] S. B. Biner, O. Buck and W. A. Spitzig, Plasticity induced crack closure in single and dual phase materials, Engineering Fracture Mechanics, 47 (1994) 1-12. [30] R. C. McClung, Finite element analysis of specimen geometry effects on fatigue crack closure, Fatigue Fracture Engineering Materials Structures, 17 (8) (1994) 861-872. [31] N. E. Ashbaugh, B. Dattaguru, M. Khobaib, T. Nicholas, R. V. Prakash, T. S. Ramamurthy, B. R. Seshadri and R. Sunder, Experimental and analytical estimates of fatigue crack closure in an aluminium-copper alloy. Part II: A finite element analysis, Fatigue and Fracture of Engineering Materials and Structures, 20 (1997) 963-974. [32] J. Z. Zhang and P. Bowen, On the finite element simulation of three-dimensional semi-circular fatigue crack growth and closure, Engineering Fracture Mechanics, 60 (1998) 341360. [33] L. W. Wei and M. N. James, A study of fatigue crack closure in polycarbonate CT specimens, Engineering Fracture Mechanics, 66 (2000) 223-242. [34] M. R. Parry, S. Syngellakis and I. Sinclair, Numerical modelling of combined roughness and plasticity induced crack closure in fatigue, Material Science and Engineering A, A291 (2000a) 224-34. [35] M. R. Parry, S. Syngellakis and I. Sinclair, Numerical, Modelling of roughness and plasticity induced crack closure effects in fatigue, Material Science Forum, 331-337 (2000b) 1473-1478. [36] M. R. Parry and S. Syngellakis, I. Sinclair Investigation of roughness induced crack closure effects in fatigue, Damage and Fracture Mechanics VI, (Editors: A.P.S Selvadurai, and C.A. Brebbia) (2000c) 313-322, WIT press, Southampton, Boston. [37] H. Andersson, C. Persson and T. Hansson, Crack growth in IN718 at high temperature, International Journal of Fatigue, 23 (2001) 817-827. [38] H. Andersson, C. Persson, T. Hansson, S. Melin and N. Jarvstrat, Onstitutive dependence in finite-element modelling of crack closure during fatigue, Fatigue and Fracture of Engineering Materials and Structures, 27 (2004) 75-87. [39] K. Solanki, S. R. Daniewicz and J. C. Newman, Finite ele-

ment modelling of plasticity-induced crack closure with emphasis on geometry and mesh refinement effects, Engineering Fracture Mechanics, 70 (2003) 1475-1489. [40] K. Solanki, S. R. Daniewicz and J. C. Newman, Finite lement analysis of plasticity-induced fatigue crack closure: an overview, Engineering Fracture Mechanics, 71 (2003a) 149-171. [41] S. Roychowdhury and Jr. R. H. Dodds, A numerical investigation of 3-D small-scale yielding fatigue crack growth, Engineering Fracture Mechanics, 70 (2003) 2363-2383. [42] S. Roychowdhury and Jr. R. H. Dodds, Three dimensional effects on fatigue crack closure in the small scale-yielding regime, Fatigue and Fracture of Engineering Materials and Structures, 8 (2003a) 663-673. [43] F. V. Antunes, L. F. P. Borrego, J. D. Costa and J. M. Ferreira, A numerical study of fatigue crack closure induced by plasticity, Fatigue and Fracture of Engineering Materials and Structures, 27 (2004) 825-835. [44] S. Kibey, H. Sehitoglu and D. A. Peckard, Modeling of fatigue crack closure in inclined and deflected cracks, International Journal of Fracture, 129 (2004) 279-308. [45] R. C. McClung, Finite element analysis of fatigue crack closure: a historical and critical review, Proceedings of Seventh International Fatigue Crack Conference, Beijing, China, 1 (1999) 495-502. [46] K. Ogura, K. Ohji and Y. Ohkubo, Fatigue crack growth under biaxial loading, International Journal of Fracture 10 (1974) 609-610. [47] D. F. Socie, Prediction of fatigue crack growth in notched members under variable amplitude loading histories, Engineering Fracture Mechanics, 9 (1977) 849-865. [48] L. Anquez and G. Baudin, Correlation between numerically predicted crack opening load and measured load history dependent crack growth threshold, ASTM STP 982 (1988) 380-397. [49] J. C. Nagtegaal, D. M. Parks and J. R. Rice, On numerically accurate finite element solutions in the fully plastic range, Computer Methods in Applied Mechanics and Engineering, 4 (1974) 153-177. [50] P. L. Lalor, H. Sehitoglo and R. C. McClung, Mechanics aspects of small crack growth from notches-the role of crack closure, In The Behaviour of Short Fatigue Cracks, EGF 1, Mechanical engineering publications, London (1986) 369379. [51] J. D. Dougherty, J. Padovan and T. S. Srivatsan, Fatigue crack propagation and closure behaviour of modified 1070 steel: finite element study, Engineering Fracture Mechanics, 56 (1997) 189-1997. [52] M. R. Parry, Finite element and analytical modelling of roughness induced fatigue crack closure, PhD thesis, University of Southampton, Southampton (2000) UK. [53] N. Kamp, M. R. Parry, K. D. Singh and I. Sinclair, Analytical and finite element modeling of roughness induced crack closure, Acta Materiala, 52 (2004) 343-353. [54] C. H. Wang, L. R. F. Rose and J. C. Newman, Closure of

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

3023

plane-strain cracks under large-scale yielding conditions, Fatigue and Fracture of Engineering Materials and Structures, 25 (2002) 127-139. [55] L. G. Zhao, J. Tong and J. Byrne, The evolution of the stress-strain fields near a fatigue crack tip and plasticityinduced crack closure revisited, Fatigue and Fracture of Engineering Materials and Structures, 27 (2004) 19-29. [56] K. D. Singh, K. H. Khor and I. Sinclair, Roughness and plasticity induced crack closure effects under single overloads: Finite element modelling, Acta Materialia, 54 (2006) 4393-4403. [57] K. D. Singh, M. R. Parry and I. Sinclair, Some issues on finite element modelling of plasticity induced crack closure due to constant amplitude loading, International Journal of Fatigue, 30 (2008) 1898-1920. [58] K. D. Singh, M. R. Parry and I. Sinclair, Finite element and analytical modelling of crack closure due to repeated overloads, Acta Materialia, 56 (2008a) 835-851. [59] Y. Li, J. He, Z. Zhang and L. Wang, Characterizing of crack closure by numerical analysis, Advanced Materials Research, 33-37 (2008) 273-278. [60] S. Pommeir and P. H. Bompard, Bauschinger effect of alloys and plasticity-induced crack closure: a finite element analysis, Fatigue and Fracture of Engineering Materials and Structures, 23 (2000) 129-139. [61] S. J. Park, Y. Y. Earmme and J. H. Song, Determination of most appropriate mesh size for a 2-D finite element analysis of fatigue crack closure behaviour, Fatigue and Fracture of Engineering Materials and Structures, 20 (1997) 533-545. [62] R. S. Blandford, S. R. Daniewicz and J. D. Skinner, Determination of the opening load for a growing fatigue crack: evaluation of experimental data reduction techniques and analytical models, Fatigue and Fracture of Engineering Materiuals and Structures, 25 (2002) 17-26. [63] D. S. Dawicke, A. F. Grandt and Jr. J. C. Newman, Three dimensional crack closure behaviour, Engineering Fracture Mechanics, 361 (1990a) 111-121. [64] J. E. Allison, Ku. R. C. Pompetzki, M. A, A comparison of measurement methods and numerical procedures for experimental characterization of fatigue crack closure, ASTM STP 982 (1988) 171-185. [65] S. K. Ray and A. F. Grandt, Comparison of methods for measuring fatigue crack closure in a thick specimen, In Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 197-213. [66] J. K. Donald, A procedure for standardising crack closure model, ASTM STP 982 (1988) 222-229. [67] J. Wu and F. Ellyin, A study of fatigue crack closure by elastic-plastic finite element analysis for constant amplitude loading, International Journal of Fracture, 82 (1996) 43-65. [68] C. M. Ward-Close and R. O. Ritchie, On the role of crack closure mechanisms in influencing fatigue crack growth following tensile overloads in a titanium alloy: Near threshold versus higher K behaviour, Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 93-111.

[69] R. C. McClung, The influence of applied stress, crack length, and stress intensity factor on crack closure, Metallurgical Transactions, 22A (1991a) 1559-1571. [70] S. R. Daniewicz and J. M. Bloom, An assessment of geometry effects on plane stress fatigue crack closure using modified strip-yield model, International Journal of Fatigue, 18 (1996) 483-490. [71] X. Zhang, A. S. L. Chang and G. A. O. Davies, Numerical simulation of fatigue crack growth under complex loading sequences, Engineering Fracture Mechanics, 42 (1992) 305321. [72] O. Nguyen, E. A. Repetto, M. Oritz, R. A. Radovitzky, A cohesive model of fatigue crack growth. International Journal of Fracture, 110 (2001) 351-369. [73] Roe, K. Siemund, T. Simulation of fatigue crack growth via a fracture process zone model, First MIT Conference on Computational Fluid and Solid Mechanics, MA, USA (2001). [74] N. Sukumar, D. L. Chopp and B. Moran, Extended finite element method and fast marching method for threedimensional fatigue crack propagation, Fatigue and Fracture of Engineering Materials and Structures, 70 (2003) 2948. [75] N. A. Fleck, Plane strain crack closure, PhD thesis, University of Cambridge, England (1984). [76] N. A. Fleck, Influence of stress state on crack growth retardation, In Basic Questions in Fatigue I, ASTM STP 924 (1988) 157-183. [77] W. Sun and H. Sehitoglu, Residual stress fields during fatigue crack growth, Fatigue and Fracture of Engineering Materials and Structures 15 (1992) 115-128. [78] R. C. McClung, B. H. Thacker and S. Roy, Finite element visualisation of fatigue crack closure in plane stress and plane strain, International Journal of Fracture, 50 (1991) 27-49. [79] S. J. Park and J. H. Song, Simulation of fatigue crack closure behaviour under variable-amplitude loading by a 2D finite element analysis based on the most appropriate mesh size concept, Advances in Fatigue Crack Closure Measurements and Analysis, ASTM STP 1343 (1999) 337-348. [80] J. C. Newman, Prediction of crack growth under variableaplitude loading in thin-sheet 2024-T3 aluminium alloys, Engineering against fatigue, University of Sheffield (1997) March 1997. [81] A. Palazotto and E. Bendnarz, A finite element investigation of viscoplastic-induced closure of short cracks at high temperatures, Fracture Mechanics: Perspectives and Directions, ASTM STP 1020 (1989) 530-547. [82] A. Palazotto and J. C. Mercer, A finite element comparison between short and long within a plastic zone due to notch, Engineering Fracture Mechanics, 35 (1990) 967-986. [83] R. W. Landgraf, Control of fatigue resistance through microstructure- ferrous alloys. In Fatigue and Microstructure, Proceeding of the 1978 ASM Materials Science Seminar, Ohio: American Society of Metals (1979) 439-466.

3024

K. D. Singh et al. / Journal of Mechanical Science and Technology 25 (12) (2011) 3015~3024

[84] W. Elber, Fatigue crack closure under cyclic tension, Engineering Fracture Mechanics, 2 (1970) 37-45. [85] J. Lankford, D. L. Davidson and K. S. Chan, The influence of crack tip plasticity in the growth of small fatigue cracks, Metallurgical Transactions, 15A (1984) 1579-1588. [86] F. O. Riemelmoser and R. Pippan, Plasticity induced crack closure under plain strain condition in terms of dislocation arrangement, Proceedings Fatigue 96, Edited by G. Lutjering and H. Nowwack (1996) 363-368. [87] F. O. Riemelmoser and R. Pippan, Discussion of error in the analysis of the wake dislocation problem, Material Transactions A, 29A (1998b) 1357-1359. [88] F. O. Riemelmoser and R. Pippan, Mechanical reasons for plasticity-induced crack closure under plane strain conditions, Fatigue and Fracture of Engineering Materials and Structures, 21 (1998a) 1425-1433. [89] B. Budiansky and J. W. Hutchinson, Analysis of closure in fatigue crack growth, Journal of Applied Mechanics, 45 (1978) 267-276. [90] H. Fuhring and T. Seeger, Dugdale crack closure analysis of fatigue cracks under constant amplitude loading, Engineering Fracture Mechanics ,11 (1979) 99-122. [91] K. Minakawa, G. Levan and A. J. McEvily, The influence of load ratio on fatigue crack growth in 7090-T6 and IN9021-T4 P/M aluminium alloys, Metallurgical Transactions, 17A (1986) 1787-1795. [92] N. Louat, K. Sadananda, M. Duesbury and A. K. Vasudevan, Theoretical evaluation of crack closure, Metallurgical Transactions, 24A (1993) 2225-2232. [93] N. A. Fleck and R. A. Smith, Crack closure-is it a surface phenomenon?, International Fracture Fatigue, 4 (1982) 157-160. [94] G. H. Bray, A. P. Reynolds and E. A. Starke, Mechanisms of fatigue crack retardation following single tensile over-

loads in powder metallurgy aluminium alloys, Metallurgical Transactions, 23A (1992) 3055-3066. [95] J. Llorca and V. Sanchez-Galvez, Dynamic analysis of plasticity-induced fatigue crack closure, Advances in Fatigue Science and Technology, Kluwer Academic Press, Denventer (1989) 809-819. [96] H. Sehitoglu and W. Sun, Modelling of plane strain fatigue crack closure, Transactions of ASME, Journal of Engineering Materials and Technology, 113 (1991) 31-40. [97] A. F. Blom, Near threshold fatigue crack growth and crack closure in 17-4 PH stell and 2024-T3 aluminium alloy, Fatigue Crack Growth Threshold Concepts, Edited by D.L. Davidson, and S. Suresh, TMS-AMIE, PA. Warrendale, (1984) 263-279. [98] M. N. James, M. N. Pacey, L. W. Wei and E. A. Patterson, Characterisation of plasticity-induced crack flank contact force versus plastic enclave, Engineering Fracture Mechanics, 70 (2003) 2473-2487. [99] J. Llorca, Roughness induced fatigue crack closure: a numerical study, Fatigue and Fracture of Engineering Materials and Structures, 15 (1992) 655-669. [100] S. Suresh and R. O. Ritchie, A geometric model for fatigue crack closure induced by fracture surface morphology, Metallurgical Transactions, 13A (1982a) 1627-1631.

Konjengbam Darunkumar Singh completed a Ph.D at Southampton University in 2005. He joined the Indian Institute of Technology Guwahati in 2006. He is engaged in several research areas of fatigue and fracture mechanics.