Until increased support is available we have postponed efforts 

 to evaluate the effects of pressure cycling and to develop a new 

 float which is both structurally and economically satisfactory. 

 In regard to the development of a new float, we have stimulated 

 some company- funded effort in a local manufacturing concern 

 and have some ideas of our own which merit exploitation. 

 Corrosion and protective coatings are discussed in a later 

 division of this section. Leakage is not a problem with any 

 float tested. Sympathetic implosion in the ocean has not been 

 observed when the floats are spaced more than one foot apart, 

 except when the depth is much greater than nominal. 



Our principal concern has been with the problem of creep 

 or plastic flow. Test data obtained on a large number of XX, 

 Special Marola 200 (1967 small lot), and Special Marola 200 

 (1968 large lot) are tabulated in Ref. 10. It is immediately 

 evident that the 1968 large lot of Marolas is grossly inferior 

 to the 1967 small lot, even though the supplier attempted to 

 manufacture the 1968 lot using the same materials and processes 

 as had been used on the 1967 lot. The reason for the discrepancy 

 has not as yet been found, and probably will not be found until 

 the precise mode of failure has been determined. Test results 

 for the 1967 lot had been very favorable, indicating that the 

 Special Marola was probably superior to the XX. At this 

 writing it appears that the Marola is superior only at pressures 

 above about 1100 PSI whereas the reverse is true below 1100 PSL 



It is believed that the XX float failure mode has been found. 

 After a period of constant pressure a flat spot develops on the 

 surface of the sphere, usually 135" (but on two occasions 45") 

 from the float handle. This flat progresses to a depresssion 

 of the surface and consequent high stresses around the depression. 

 Finally the nnaterial ruptures around the depression and a 

 massive collapse occurs. The flat or depression corresponds to 

 an anode for the first buckling mode shape. Its location 45" 

 or 135° from the handle is expected when one coisiders that the 

 float handle provides stiffening and should therefore be located 

 at a node where no change of curvature occurs. Growth of 

 the depression appears to be a creep or plastic flow process. 

 It is felt that the Marolas also fail by a creep process. Con- 

 sequently we have attempted to correlate our data with the 

 Larson-Miller chemical-mechanical equation of state which 

 expresses the equivalence of temperature and time in the creep 

 process: 



K = Tdog^Qt + 20)/ (1 - T/T^) 



T = temperature in F Absolute 



T^ = temperature in F Absolute at which strengthM) 



t = time to failure in hours 



222 



