The properties of glass pipes witii conical end flanges differ 

 substantially from those assumed in the analytical expression. The pipes, 

 first of all, are not of uniform thickness and do not possess a uniform radius. 

 This is shown by the variation in thickness of plus or minus 20% from nomi- 

 nal value and the plus or minus 4% variation of external radius measured on 

 glass pipes tested. In addition, the conical end flanges do not give the glass 

 cylinder simple end support, but a support which is neither simple nor rigid, 

 but a cross between the two. The taper in the end flanges introduces an 

 additional uncertainty: the cylinder length between end supports. If the 

 overall length of pipe sections (including the tapered flange portion) is used 

 in the equation, different critical pressures will be obtained from the equa- 

 tion than if the length of the cylindrical portion between tapered sections 

 of the pipe is used. In addition to these uncertainties, there is added the 

 presence of a stress raiser in the form of an 0-ring groove in the flange that 

 may cause the failure of the glass pipe at lower pressure than if the implo- 

 sion of the pipe took place due to elastic buckling. Only after the user of 

 the analytical expression understands all of these differences between the 

 assumptions on which the equation is based and the measurements of the 

 actual test specimens is he ready to intelligently apply the equation to the 

 calculation of critical pressure due to buckling for glass pipes with conical 

 flanges. 



For the purpose of comparing calculated with experimental implosion 

 pressures, the following dimensions and properties of pipes were used for 

 plotting of experimental results (Figure 41): 



D— mean diameter, as determined by subtracting one nominal wall 

 thickness from nominal external diameter specified by manu- 

 facturer 



t— nominal wall thickness specified by manufacturer of glass pipe 



E— nominal modulus of elasticity for borosilicate glass used in the 

 fabrication of pipe, 9.1 x 10^ psi 



L— nominal length of pipe, the distance between the two flat bearing 

 surfaces on the ends of the pipe 



Comparison of experimentally determined implosion values with the 

 calculated curves for the same t/D ratio discloses rather good agreement 

 between them. The agreement becomes more remarkable when it is noted 

 that there are so many dimensional variations in the pipes and the flanges at 

 the end of the pipes do not represent the simple supports specified by the 

 analytical expression. Only one pipe size, the 6-inch-ID x 6-inch-long pipe, 

 imploded at pressures significantly lower than the calculated pressure. 



43 



