Reliability Engineering Snapshot TM

Illustrated Case Studies in the Industrial World of Failure Analysis, Non Destructive Evaluation, and Predictive Maintenance

Machine Design

Case No. 84: January 2001

The old saying that "more is better" doesn't always hold true. As a matter of fact, where welding is concerned, there is a British standard that addresses the issue of exactly how to get the most out of a weld by optimizing its configuration. It seems that the "Brits" have figured it out that "more isn't exactly better." The following excerpts are from a study that was conducted for me by Applied Reliability Inc. This work is part of the ongoing failure analysis and design improvement work. For a full history of the failure analysis, the reader is encouraged to view the ASM technical paper whose page link is on the home page top bar. INTRODUCTION The planned installation of reinforcement rings on a rotary dryer required significant welding of the rings to the dryer shell. The optimum weld size was desired, where optimum was defined as the size beyond which there was no improvement in the system reliability. The fatigue analysis was based on British Standard BS 7608 Code of Practice for Fatigue Design and Assessment of Steel Structures. A finite element stress analysis and fatigue analysis was conducted to identify the optimum weld size. ANALYSIS The weld will fail in fatigue cracking, if sufficiently stressed, at one of two locations. These locations are the root of the weld, with cracking in the weld, and the toe of the weld, with cracking in the shell. These two failure locations can be characterized as a chain of two links. In a chain, there is little improvement in chain strength when one link is stronger than the other link. Similarly, the reinforcing ring weld factor-of-safety for cracking in the root need be no greater than that for cracking at the toe. This is especially the case here because cracking in the shell is the least desirable failure mode. The fatigue analysis, using finite element analysis, was based upon British Standard BS 7608. This standard is based upon fatigue testing of weldments and the fatigue endurance data has considered the stress concentrations associated with the weld details. However, the stress concentrations due to the geometry considered by the standard must be correctly modelled such that the modelling technique does not inject misleading information. The two dimensional finite element model used for the stress analysis is shown at the right. Symmetry was used so that only the right side of the stiffening ring is shown. The left side would be a mirror image and appropriate boundary conditions allow the complete cross section to be modeled with only half of the cross section. The root of the weld was modeled so as to eliminate the stress concentrations due to the minute weld details while maintaining the effect of weld geometry. Stress at the toe was obtained by projection of the stress adjacent to the toe, to the toe location. This avoided the stress concentration due to the weld detail at the toe, but included the stress concentration effect of geometry according to the British standard.
Coupled forces were applied to place a moment on the stiffening ring weld. Ratios of fatigue strength to stress for various weld sizes were obtained. By iteration, the weld size that provided a strength to stress ratio at the root, equal to or a little greater than at the toe, was selected (highlighted). Note that stresses, and strength to stress ratios, are relative and are not absolute values.	Weld Size (inches)	Weld Toe Stress (VM psi)	Weld Root Stress (VM psi)	Toe Ratio	Root Ratio
	1/2	2034.8	1507.8	2.85	2.40
	3/4	2091.3	1325.4	2.77	2.73
	7/8	2085.6	1302.7	2.78	2.78
	1	2083.8	1066.3	2.78	3.40

The analysis done by Applied Reliability showed some interesting information. At a 1/2" thick fillet weld, the toe was stronger than the root. This meant that even if the stress levels were acceptable, a fatigue crack would start at the root. The crack would more than likely progress up through the throat of the weld and remain undetected by any visual inspection. One would not know that the crack was there until it was too late (when using visual inspection). On the other extreme was the 1" thick fillet weld. A crack would certainly start at the toe, a desirable circumstance for visual inspection. However, it would start cracking much sooner than at the root. This would mean that a lot of weld metal would be wasted. Even worse, the amount of time to weld a 1" fillet as compared to a 7/8" fillet was considerable. All downtime required for this welding meant lost profits, so it was important to minimize downtime by picking a weld that got us the strength we needed without going overboard on the welding.