Int. J. Fatigue Vol. 20, No. 10, pp. 767773, 1998 1998. Published by Elsevier Science Ltd.

. All rights reserved Printed in Great Britain 01421123/98/$see front matter

PII: S0142-1123(98)00051-6

Stress concentration and fatigue of proled reinforcing steels

H. Zheng* and A. Abel
Department of Civil Engineering, The University of Sydney, Sydney, NSW 2006, Australia (Received 6 November 1997; revised 8 June 1998; accepted 12 June 1998)
Stress concentrations arise from proles of ribbed reinforcing steel bars and in this study the results of calculated stress concentration factors (SCF), by using nite element method, are related to the fatigue test results. It is apparent that the degree of connement of a ribbed bar embedded in concrete not only affects the magnitude of the ultimate bond stress but also the bond stress-slip relation. Thus the rib geometry or prole pattern is designed for optimum bond characteristics and not for fatigue considerations. However the rib geometry inuences the fatigue performance through the SCFs arising from the root radius, width and ank angles of the proles. It is shown that these latter factors have a signicant inuence on the fatigue behaviour of reinforcing steel bars. 1998. Published by Elsevier Science Ltd. All rights reserved
(Keywords: fatigue; reinforcing steels; stress concentration)

INTRODUCTION MacGregor et al. and Helgason et al. demonstrated that the fatigue strengths of proled reinforcing steels, bars with extrusions on the surface as shown in Figure 1, are much lower than those of smooth bars machined from the same steels. Bars with different types of proles3 and bars produced from milling rolls suffered different wearing conditions4 have shown that the radii at the bases of transverse lugs have a major inuence on fatigue strength. It was also suggested that the inuence of lug geometry on fatigue properties even offset the effect related to the grade of steels1,4,5. Although it has been attempted to quantitatively relate fatigue behaviour of reinforcing steels to their prole geometry2,5, so far there is no method which can be applied universally. This is partially due to the various SCFs arising from the rib geometry. Jhamb and MacGregor6 in 1974 carried out a theoretical stress analysis using nite element method on the stress concentration at the vicinity of lug root and concluded that the concentration factor increases as lug radius decreases and as the lug width and ank angle increase. In this paper, using a similar method, the inuences of the lug geometry parameters on stress concentration are investigated and the results are related to the appropriate fatigue test results.
1 2

PROFILE PATTERNS AND ANALYTICAL MODELS Pictorial exposition of typical prole patterns are illustrated in Figure 1 and a typical geometrical lug prole obtained by laser displacement meter is shown in Figure 2 with the parameters described below. r/h: W/h: : D/h: l/h: ratio of lug root radius to lug height; ratio of lug width to lug height; ank angle of the lug; ratio of bar diameter to lug height; the distance between neighbouring lug roots.

*Corresponding author.

Although optical comparator, plastercasting and microscopy were also used in the determination of the lug parameters, the laser displacement meter method, enables one to carry out multiple section measurements and a further advantage of this method is that the data can be analysed by computer. In order to cover a large number of lug proles, the following parameter ranges were used: r/h = 0.11.5; W/h = 3.04.0; = 4565; D/h = 816. Two models were used to simulate transverse lugs in the study of stress concentration. Model I simulates an equally spaced transverse lug pattern where each lug is relatively far away from its neighbours, Figure 3a. This model corresponds to the bars marked b, c, d and e in Figure 1. Model II simulates any unequally spaced lugs distributed in such a way that a lug is closer to its neighbour lug on one side but far away
767

768

H. Zheng and A. Abel

Figure 1

Typical prole patterns of reinforcing steels

Figure 3 Models used to simulate the geometry of transverse lugs; (a) Model Ifor long distance spaced transverse lugs; (b) Model IIfor unevenly spaced lugs

from the one on the other side, Figure 3b. This model corresponds to the bars marked a and f in Figure 1. The same models are applied in the analysis of identication marks. In the analysis the material was assumed to be elastic and isotropic with a uniform axi-symmetrical tensile loading. Due to symmetry only a quarter of each geometry pattern was analysed with boundary conditions in such a way that the displacement of nodes along ad in axial direction and those along dc in radial direction were set to zero. The mesh of Model I, consisting of 2150 8-node biquardratic (CPE8) elements, is shown in Figure 4. Elements along the circumferential direction of the lug root is xed to be 10 and thus the smaller is the lug root size the higher is the element density around the corner. Model II mesh is basically a copy of Model I. The calculation was performed employing the ABAQUS 5.57commercial program. The largest principal stresses at the surface and layers below the surface were calculated with a sequence incremental

of l0, l0, l0/ , l0/ , l0/ 2, l0/ 2 where l0 = 0.002916 h and = 0.8 and 0.7 for r = 0.3 h and r = 0.4 h, respectively. STRESS CONCENTRATIONS RELATED TO PROFILE GEOMETRY Model I A total of 135 cases with different geometry parameters were studied and the results of maximum SCFs at the surface and at the point 0.013 h below surface are given in Tables 1 and 2, respectively. Figure 5 shows typical stress concentrations along the bar axis at different depths below the surface. It is quite clear that stress concentrations are limited in a very small area in the vicinity of the lug root and the maximum stress is at the lug root junction with the base material. On the surface and along the longitudinal direction the SCF drops to 50% of the peak value at a distance of 0.01 h from the lug root and it becomes negligible at

Figure 2

Lug prole and the geometry of a transverse lug measured using laser displacement meter

Stress concentration and fatigue of proled reinforcing steels

769

Figure 4

Mesh for Model I

0.5 h from lug root. Beyond a distance of 2.5 h there is essentially no stress concentration. The lug itself does not carry large stress and at the top surface of the lug there is no stress at all. As fatigue cracks start at the bar surface, at the maximum SCF, only the maximum SCF is taken into account when fatigue is discussed. The gradient of SCF along the radial direction of the bar is rather large and increases as the r/h ratio decreases. It is shown in Figure 5 that 0.013 h below the surface for the r/h ratios of 0.5 and 0.1 the SCF drops 6.6% and 24.7%, respectively. Calculations show that the maximum value of SCF decreases only 0.22% as the bar diameter versus lug height ratio, D/h, increases from 5.2 to 28. This trend indicates that the stress concentration is independent from bar size within the practical D/h range of 8 to 16. This is in agreement with the result obtained by Jhamb and MacGregor6. The inuences of the lug root radius on SCF are presented in Figure 6 showing the great inuencing effect of the r/h ratio. Regression analysis shows that a polynomial function of an order of 6, as shown in Appendix A, is more appropriate to describe such relationship than the logarithmic function presented by Jhamb6. Some of the differences between the present results and those reported by Jhamb may be attributed to the analysis of the latter where the average SCF
W/h = 3.0 = 55 3.039 2.429 2.154 1.996 1.886 1.804 1.739 1.691 1.650 1.616 1.587 1.561 1.539 1.519 1.501 W/h = 3.0 = 65 3.072 2.449 2.174 2.015 1.902 1.818 1.761 1.711 1.670 1.635 1.605 1.580 1.559 1.541 1.524 W/h = 3.5 = 45 3.093 2.496 2.218 2.053 1.931 1.843 1.778 1.722 1.676 1.639 1.607 1.580 1.557 1.536 1.517 W/h = 3.5 = 55 3.194 2.543 2.246 2.073 1.955 1.868 1.796 1.743 1.699 1.663 1.631 1.603 1.579 1.557 1.538 W/h = 3.5 = 65 3.234 2.565 2.265 2.095 1.973 1.883 1.818 1.764 1.719 1.681 1.649 1.621 1.596 1.575 1.557 W/h = 3.0 = 45 3.215 2.589 2.296 2.123 1.996 1.900 1.829 1.771 1.725 1.685 1.651 1.622 1.595 1.572 1.551 W/h = 4.0 = 55 3.327 2.642 2.328 2.144 2.016 1.924 1.847 1.789 1.743 1.704 1.670 1.641 1.615 1.592 1.571 W/h = 4.0 = 65 3.373 2.665 2.345 2.163 2.035 1.939 1.867 1.810 1.762 1.722 1.688 1.657 1.631 1.607 1.586

Table 1 r/h 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5

Maximum stress concentration factors at the surface of lug root W/h = 3.0 = 45 2.953 2.389 2.129 1.972 1.861 1.780 1.717 1.667 1.625 1.592 1.564 1.539 1.517 1.497 1.480

Table 2 r/h 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5

Maximum stress concentration factors at the point 0.013h below surface W/h = 3.0 = 45 2.376 2.127 1.964 1.820 1.745 1.689 1.640 1.599 1.568 1.542 1.518 1.498 1.479 1.463 1.448 W/h = 3.0 = 55 2.446 2.159 1.989 1.847 1.771 1.711 1.663 1.625 1.593 1.565 1.541 1.520 1.501 1.584 1.469 W/h = 3.0 = 65 2.472 2.183 2.010 1.865 1.786 1.725 1.684 1.645 1.612 1.584 1.559 1.538 1.518 1.503 1.491 W/h = 3.5 = 45 2.480 2.218 2.045 1.892 1.808 1.747 1.698 1.653 1.615 1.586 1.561 1.538 1.518 1.500 1.484 W/h = 3.5 = 55 2.561 2.257 2.071 1.918 1.836 1.771 1.717 1.675 1.640 1.610 1.584 1.561 1.540 1.522 1.505 W/h = 3.5 = 65 2.595 2.283 2.094 1.937 1.852 1.785 1.738 1.695 1.659 1.628 1.602 1.578 1.557 1.538 1.521 W/h = 3.0 = 45 2.569 2.297 2.116 1.955 1.866 1.799 1.746 1.702 1.664 1.631 1.603 1.578 1.555 1.535 1.516 W/h = 4.0 = 55 2.659 2.341 2.142 1.980 1.893 1.824 1.764 1.720 1.682 1.650 1.622 1.597 1.575 1.555 1.537 W/h = 4.0 = 65 2.699 2.368 2.167 2.000 1.909 1.838 1.785 1.739 1.701 1.668 1.639 1.613 1.590 1.570 1.552

770

H. Zheng and A. Abel

Figure 5

Analytical results-stress distributions along longitudinal axis

Figure 6

Stress concentration factor changes as a function of r/h

values were taken at 0.026 h below surface. It is believed that the SCF at the surface is more important than the average value at a certain depth when fatigue evaluation takes place since cracks develop from the plastic deformation of crystals at the very surface. Generally, the SCF increases as the lug width and ank angle increase, but these inuences are much smaller than that of the r/h ratio as shown in Figure 7. The data in Table 1 indicates that the change in the value of W/h by 0.5 leads to no more than 5.3% change in SCF, while the change in ank angle by 15 degree results in 3.5% change in SCF at most. In addition, these effects decrease as r/h increases and the effects of ank angle becomes negligible when r/h > 0.5.

Model II This analysis shows that as the distance between the two neighbouring lugs diminishes the SCF increases at the inner side of the lugs, Figure 8. This increase in the value of SCF reaches a maximum of 2.65 when the two lugs merge. On the outside of the lug, the maximum increase in the SCF values is only 2%. In fatigue test the cracks developed from these high SCF points, Figure 9. DISCUSSION The high SCF resulted in large number of fatigue crack initiations along the lug root as shown in Figure 10. After initiation the cracks propagated away from

Stress concentration and fatigue of proled reinforcing steels

771

Figure 7

Stress concentration variation with lug widths, ank angles and r/h ratios

Figure 8

Analytical results of Model II

Figure 9

Cracks initiated at the intersection point where the SCF is the largest

the transverse lug on the bar surface into the body of the bar normal to the bar axis. This no doubt suggests that the largest stresses lie in the longitudinal direction, as otherwise the cracks would have grown along the root where the magnitudes of stress concentrations are much higher than elsewhere. In fact, Jhamb6 measured that the strain in the longitudinal direction was always higher than that in the direction normal to the lug at the root. These results and models used for the analysis are therefore compatible.

According to the above analysis sharper merging of a prole into the base of the bar substantially increases the SCF and therefore reduces fatigue crack initiation time. Identication marks, indicating the name of the supplier, prove this point strongly, Figure 11. The radius of the merging root of the identication mark is nearly zero while its height is of the order of 88% of the value measured for the height of the transverse lugs. Thus bars containing identication marks failed from identication mark roots, without exception, and

772

H. Zheng and A. Abel

Figure 12 Crack initiated from a sharp identication mark close to a transverse lug

Figure 10 Crack initiation in an evenly spaced transverse lug

Figure 13 Crack initiation at the identication mark where is a small gap between the mark and lug Figure 11 Crack initiated from a sharp identication mark

exhibited fatigue limits far lower than the bars free of such marks. Figures 12 and 13 also show fatigue crack initiations associated with identication marks. Even when identication marks and ribs are in close proximity, which increases the SCFs according to Model II, the identication mark appears to be the crack initiator. In Table 3 the sharpest geometry parameters are listed for seven TEMPCORE reinforcing bars together with the calculated SCFs and measured fatigue limits. The data in the table indicates a good correlation: when the r/h increases from 0.3 (32 and 36 mm bars)

to 1.0 (16 and 20 mm bars) the SCFs decrease approximately by 23% and the fatigue limit increases by about 30%. Although the general trend is that the fatigue limit decreases when the calculated SCF increases, there is some scatter if the fatigue limit is plotted against SCF. This indicates that the fatigue limit is not dominated entirely by SCFs but is inuenced by residual stresses and microstructure features. It may also suggest that in the calculation of SCF the surface roughness should be taken into account. The cracks shown in Figure 10 are not going through the lugs indicating the validity of the analytical results,

Stress concentration and fatigue of proled reinforcing steels

Table 3 Geometry, stress concentration factors and fatigue strengths of TEMPCORE reinforcing bars Bar size 12 mm 16 mm 20 mm 24 mm 28 mm 32 mm 36 mm 36.0 42.0 44.3 46.7 41.7 37.5 39.4 r/h 0.9 1.08 0.96 0.15 0.48 0.31 0.30 W/h 4.0 4.20 4.07 2.79 3.4 3.65 3.7 Calculated SCF 1.700 1.665 1.704 2.562 1.930 2.174 2.216 Fatigue limit 270 MPa 310 MPa 310 MPa 240 MPa 230 MPa 240 MPa 240 MPa

773

6. Fatigue initiation frequently takes place at identication marks as they sharply merge into the bar base and also at neighbouring transverse lugs when they are in close proximity.

REFERENCES
1 2 MacGregor, J. G., Jhamb, I. C. and Nuttall, N. ACI Journal, 1971, March, 169. Helgason, T., Hanson, J. M., Somes, N. F., Corley, W. G. and Hognestad, E. Fatigue strength of high-yield reinforcing bars. National cooperative highway research program report 164, Transportation research board national research council, 1976, Washington, D. C. Hanson, J. M., Burton, K. T. and Hognestad, E, Journal of PCA research and development laboratories, 1968, September, 2. Gronqvist, N.-O. In Concrete Bridge Design ACI SP26-38, Amer ican Concrete Institute, Detroit, 1971, p. 1011. Jhamb, I. C. and MacGregor J. G. In Fatigue of Concrete, ACI SP publication 41-7, 1973, p. 139. Jhamb, I. C., and MacGregor J. G. In Fatigue of Concrete, ACI publication SP 41-8, American Concrete Institute, Detroit, 1974, p. 169. ABAQUS 5.5. Hibbitt, Karlsson and Sorensen, Inc. 1995.

that is, the stress concentration on the lug surface and in the body of the lug are not signicant. In cases where the transverse lugs are normal to the bar axis, however, cracks initiate in one plane of the cross section and coalesce easily. Thus inclined transverse lug pattern is better in terms of fatigue resistance. CONCLUSION From the nite element approach and the obtained experimental data the following conclusions may be drawn. 1. The proposed Models can predict SCFs associated with reinforcing bar prole patterns. 2. The root radius is the prime factor affecting the value of SCF. Smaller root radius naturally results in a higher stress concentration factor. 3. Stress concentration increases as the prole width and ank angle increase but these inuences are much smaller than those relating to the r/h ratio. The effects of deformation width are more signicant than those of ank angles. 4. Transverse lugs normal to the bar axis reduce fatigue life when compared with bars where the lugs are angled to the bar axis. 5. Stress concentration increases as the lug spacing approaches zero.

3 4 5 6 7

APPENDIX A Regression results for the relationship between SCF and the r/h. W/h = 4.0; = 65SCF = 5.9972(r/h)6 32.812(r/h)5 + 71.994(r/h)4 81.191(r/h)3 + 50.306(r/h)2 17.241(r/h) + 4.6629; W/h = 4.0; = 45SCF = 5.1408(r/h)6 28.031(r/h)5 + 61.293(r/h)4 68.951(r/h)3 + 42.799(r/h)2 14.906(r/h) + 4.3359; W/h = 3.0; = 65SCF = 5.1524(r/h)6 28.274(r/h)5 + 62.262(r/h)4 70.497(r/h)3 + 43.855(r/h)2 15.067(r/h) + 4.1992; W/h = 3.0; = 45SCF = 4.6015(r/h)6 25.153(r/h)5 + 55.143(r/h)4 62.191(r/h)3 + 38.679(r/h)2 13.455(r/h) + 3.9643