Page:NTSB RAR-73-5.pdf/55

APPENDIX G depth of the bevel did not provide access to permit welding over the full ½-inch thickness. In addition, only 25-percent fusion was achieved in the accessible area.

The possibility of not achieving a good weld with full penetration was quickly determined during the construction of the cars by the production department of St. Louis Car Division. The bevel was reduced in depth and an additional weld adjacent to the weld in the beveled area was added in an effort to compensate for any deficiencies of penetration. This effective design change was not, however, changed on the drawing. These factors are illustrated by the attached sketches based upon exhibits in evidence. The added weld is visible in this exhibit.

The added weld did not solve the problem. First, as can be seen in the sketch, the added weld material was distant from the location at which it could add to the strength of the weakest section. This section was limited by the 5/16-inch bevel. Second, the poor fusion at the weakest section could not be corrected by metal added to the outside of the plate.

Fusion to the full depth of the plate was assumed in the calculations for the collision posts based on the original drawing. Any reduction of cross section, such as that produced by the smaller bevel or poor fusion, would reduce the shear strength of the collision post to below the required 300,000 pounds.

It also appears that the welding design for the attachment of the collision post to the underframe relied upon assumptions not justified by documents of the current engineering practice. Specifically, the weld strength calculations made in allegedly meeting the Federal requirements assumed ideal welding conditions, whereas weaknesses of the actual conditions had been warned against in the American Welding Society's Welding Handbook, Sixth Edition, Section One, Paragraph 8.14. This paragraph states that joints made with single-bevel welds from one side:

"1. Are difficult to obtain a sound weld due to one perpendicular groove face. Vee and U grooves are preferred.

2. * * * should not be used when tension due to bending is concentrated at the root of the weld or when subject to fatigue, impact loading or service at low temperature.

4. Strength depends on degree of joint penetration, which is usually less than the depth of chamfering. * * *" 993C