A procedure for fatigue design of web-stiffened cruciform connections

Abstract

Fatigue design of dynamic loaded ship structures is normally based on finite element analysis using shell elements where the welds are not included in the analysis models. Finite element analyses of a fatigue-tested hopper knuckle type specimen were performed for calibration of analysis methodology in a joint industry project. The test specimen was analyzed using solid and shell elements. Only the solid element models included the weld geometry. In the present work, the finite element analyses were extended to other geometries at web-stiffened cruciform joints such as typical found at hopper connections, at stringer heels, and at joints connecting deck structures to vertical members in ship structures. Based on the result from these analyses, a methodology for derivation of hot spot stress at welded connections accounting for the weld size using shell finite element models was developed. The basis for this procedure is presented in this paper.

Study on the preciseness of hot spot stress of web-stiffened cruciform welded joints derived from shell finite element analyses

“…This suggests that it is better to calculate s s (c shift ) by means of cubic polynomial interpolation in HSS Lotsberg calculations. Table 2 and Table 3 show that the estimate ratios of HSS Lotsberg (cubic interpolation) are slightly larger The original calibration analyses of Lotsberg et al [9,10] were related to weld joint models with a limited flange width subject to simple bending loads. Lotsberg et al [10] carried out additional analyses to investigate the effects of the increased plate width and loading types (pure membrane stress in the flange plate and pressure on the flange plate), and they reported that the results for different types of loading and with larger plate width did not deviate much from that of the original calibration.…”

“…Table 2 and Table 3 show that the estimate ratios of HSS Lotsberg (cubic interpolation) are slightly larger The original calibration analyses of Lotsberg et al [9,10] were related to weld joint models with a limited flange width subject to simple bending loads. Lotsberg et al [10] carried out additional analyses to investigate the effects of the increased plate width and loading types (pure membrane stress in the flange plate and pressure on the flange plate), and they reported that the results for different types of loading and with larger plate width did not deviate much from that of the original calibration. However, the type of loading and model's structure in these additional analyses are simpler than those of full-scale models under the real load examined in this study, and there is concern that there are some cases where the effects of pressure loading and multi axial stress field reduce the accuracy of Lotsberg's method substantially.…”

“…Lotsberg's method [9,10] gives the formula for limited structural geometry (f is limited to 90, 60 and 45 deg.). The correction factor b for arbitrary bevel angle can be determined by interpolating those for given angles, but there is as yet no validation study for such procedure.…”

“…HSSs are derived from shell FE results following the author's proposal (the procedure described in Chapter 3), Lotsberg's method (Lotsberg et al [9,10]) and conventional 0.5tÀ1.5t linear extrapolation 1200.0 1400.0 1400.0 1400.0 For conventional 0.5t-1.5t extrapolation, the shell element stresses are read out at points shifted away from the intersection by 0.5t and 1.5t. It is not needed to interpolate element stresses because t  t meshes are used in shell analyses.…”

“…Lotsberg et al [9] [10] proposed yet another determination method. The authors (Osawa et al [8]) examined the validity of these two methods, and they demonstrated that these 'new generation' methods give almost identical HSS, and those methods reduce the excessive safety allowance of HSS determined by the conventional 0.5t-1.5t extrapolation.…”