Design Procedure Review – 02, Transverse Fillet Weld by Kotur Raghavan
It would be hard to come across structures and mechanical assemblies which do not contain welded connections. However, in almost all the stress analysis tasks using FEA the weld geometries are not explicitly accounted for. Components which are welded together are assumed to be rigidly connected. One of reasons is the challenges involved because of geometric complexities. Moreover, welded connections are normally configured based on industry standards.
All the same, we find chapters on design of welds in all books of structural and mechanical design. One such problem is taken up for study in the present article. This is the second in the DPR series of articles.
Transverse Fillet Weld
The configuration taken up for study is shown in Fig. 1 below. Load transfer takes place between the plate 1 and plates 2 through the welds 3.
The present study is in the context of the design procedure described by Shigley and Minschke in their book Mechanical Engineering Design, Sixth edition, Page 529.
The authors of the book have used the free body diagram shown in Fig. 2 above. The components of the force are the normal force N and the shear force S acting on the throat section as shown. These forces on the throat section are assumed to be generating uniform stress fields. Through solution of equilibrium equations, they have derived an important design relation for the maximum von Mises stress Se as
Where L is the length of the plates and weld.
Finite element model study is taken up for assessment of the aforementioned design information. 2D plane stress model is analysed. Because of symmetry only one half of the assembly is considered for analysis. The details of the model are shown in Fig. 3 below.
The allowable stress of the weld material is assumed to be 200 MPa. For this allowable value and a thickness of 20 mm the load limit for the weld works out 1852 newtons per unit length. This load is applied in the form of uniform pressure (92.6 MPa).
The overall results are presented in Fig. 4 above. The high stress is at the corner of the weld and is due to singularity. As the region of interest is the weldment, we will focus on the prevailing stresses there.
The weld area and the equivalent stress contours in the region are given in Fig. 5 above. The range is from zero to the allowable stress (200). The grey coloured regions have stresses higher than the allowable. The high stress regions also include spurious stresses due to singularity. As the throat area is critical, it is necessary to assess the nominal value of stress at the corner A. One way of addressing this task is to compute linearized stresses across the section AB.
In Fig. 6 above, linearized von Mises stresses across the section AB are shown. The graph has two important results. The section under consideration has a membrane component of 173 MPa and a Membrane + bending component of 309 MPa. The value of 309 is the nominal stress at the corner A. This value is still high in comparison with the allowable value of 200. At this point we will take a look into ASME’s Design by Analysis (DBA) guidelines.
DBA guidelines are part of the ASME’s Boiler and Pressure Vessel Codes, Section VIII, Division 2. As per this, stresses across any section can be categorized into membrane (uniform component), bending (linearly varying component) and the Peak (additional due to stress raisers). From a strength point of view, as per the code, the allowable stress, which is normally the yield stress divided by the factor of safety, applies to the membrane component. The allowable limit for membrane + bending component is 1.5 times that for the membrane component.
The present author has earlier stated that the machine design books and the mode of teaching make no reference whatsoever to the stress categories. See
This shortcoming results in conservative designs.
In the present case study, the allowable membrane stress is 200 and therefore the allowable (M + B) component of stress is 300 MPa. The computed membrane stress is 173 which is well within the allowable limit. The (M + B) component is 309 and is thus marginally unsafe (by 3 per cent). If the higher allowable is not considered the design would have been unsafe by a big factor.
Stresses along the throat
As stated already, the normal as well as the shear stress at the throat section are assumed to be uniform in the design using hand calculations.
In Fig. 7 above, the computed values of normal and shear stresses are plotted along AB. The two corresponding curves are named Normal-FE and Shear-FE in the graph. The two other lines, in broken lines, show the values of these as assumed in the Machine Design (MD) text book. The shear stress distributions are in reasonable agreement. Also the assumed value is higher than the computed value for most part. The assumed value of normal stress is lower than the computed value for more than 50 per cent of the path. This is the main reason for the design assumption turning out to be non-conservative.
· The design procedure, as presented in the referred text book, has been thoroughly reviewed.
· The assumption of uniform stress distribution along the throat section has turned out to be non-conservative. The nominal value of (M+B) component is observed to be more than one and half times the allowable value (309 as against 200)
· However, if the DBA guidelines are considered, the design will be only marginally unsafe (309 as against 300).
· Whether the DBA guidelines are taken into consideration or not, the fact is that the weld is acceptable up to about 97 per cent of the applied load. Thus the conservatism involved in treating the (M+B) component on par with the membrane component has almost fully compensated for the non-conservative assumption of uniform stress distribution. This is sort of ironical situation.
· An important takeaway of this study is the use of stress linearization and its application for certifying the design adequacy. In general, linearization is possible only through finite element simulation.