On the growth of small fatigue cracks in MIG welded automotive grade steels
PadayatchyYogen
2017
Fabrication by welding is very popular in the automotive industry due to its effectiveness in reducing production costs and it has been found that failures in those welded components can arise due to fatigue cracking caused by cyclic loading. Therefore, the ability to accurately predict the time for a small non-detectable flaw to reach a size where it is first detectable is important because of the resultant potential to reduce component validation test programs. However, when considering how to best predict crack growth it should be noted that the science of fatigue crack growth has traditionally revolved around the relationship between stress intensity factor range, ΔK, and crack growth rate, da/dN, which is commonly believed to have three distinct regions whereby Region I is associated with crack growth at low ∆K’s and generally accounts for a significant proportion of the fatigue life of a structure. It is this region that is important if we are to determine the time for a small non-detectable flaw to grow to the size when it is first detectable.
The current approach in the automotive industry is to design against crack growth and this is generally done via the use of stress vs. number of cycles (S-N) curves or strain life methods. Most existing fracture mechanics approaches used to assess the fatigue performance in MIG welded automotive structures are based on the similitude hypotheses which means that two different cracks growing in identical materials with the same thickness, with same stress intensity factor range DK, and the same maximum stress intensity factor, Kmax, will grow at the same rate. However, using the principle of similitude to predict the growth of short cracks can lead to errors as short cracks grow faster than long cracks for the same values of stress intensity factor, and they can grow below the threshold stress intensity factor of long cracks. This phenomenon is known as the ‘short crack effect’.
This research in this thesis focusses on investigating the applicability of two non-similitude crack growth equations namely the Generalised Frost-Dugdale (GFD) equation and the Hartman-Schijve-McEvily (HSM) equation, on the fatigue crack growth of small cracks in MIG welded automotive grade steels.
A fatigue test program in MIG welded Xtraform 400 (XF400) and HA350 automotive grade steel V-notched SENT specimens was performed, for a range of R-ratios, to generate experimental short crack data. The results revealed that, in all cases crack growth was confined to the weld and the cracks on each side of a sample generally grew as individual cracks before joining and growing as a through crack.
A 3D finite element technique has also been successfully developed and validated, in demonstration of a numerical procedure that can be efficiently used together with experimental results to obtain stress intensity factor solutions to be used in the GFD and HSM equations to assess their capabilities of predicting fatigue crack growth in MIG welded automotive grade steel components.
The applicability of the GFD crack growth equation was investigated for the growth of short cracks (cracks typically <1.5 mm in this research was considered as short crack). It was found that crack growth was not a unique function of DK as previously reported by other researchers but instead could be represented as a function of ∆K, Kmax and crack length. This finding implies that similitude based crack growth equations such as Paris equation and its variants, which are commonly used in the automotive industry, are questionable when used to predict fatigue life of short cracks in automotive grade steels. However, the disadvantage of this method is that the value of p in the GFD crack growth equation appears to be R-ratio dependent for welded structures and as such this approach is not recommended.
The applicability of the HSM crack growth equation (a variant of the Nasgro equation) was also investigated for short cracks in MIG welded automotive grade steels XF400 and HA350. It was seen that the crack growth data can often be reasonably well represented with an exponent α that is approximately 2 (conforming to previous researchers who have presented crack growth equations whereby da/dN is related to (∆K - ∆Kth)² and a material constant, D of approximately 5.5 x 10 10. For rail and aerospace materials, the value of ∆Kthr asymptotes to zero as the crack length decreases whereas the residual stresses associated with the welding process mean that that this is not true for welded structures. In this case, it appears that the effect of the residual stresses resulted in a non-zero value of ∆Kthr. This finding suggests that the Hartman-Schijve-McEvily variant of the Nasgro equation is superior when attempting to model crack growth in automotive welds and thereby presents an alternative to computing the time to grow a crack from a small non detectable size to a size that can be visually detected and hence an alternative to S-N based design curves.