catastrophic explosion failure giving the windows a proven safety factor 
of 5 under a single sustained long-term overload (approximately 10 
minutes duration). 
The mode of failure under long-term loading was found to be similar 
to the mode of failure under short-term loading and thus will not be 
discussed here in any detail. There was, however, a significant differ- 
ence in the magnitude of window deformation prior to catastrophic failure. 
While under short-term loading the maximum displacement of the 1.040- 
thick window's center just prior to failure was approximately 0.250 to 
0.350 inches, for long-term loading the displacement was 0.400 to 0.500 
inches (Figure 9). Surprisingly enough, the maximum displacement prior 
to catastrophic failure under long-term loading was the same regardless 
of the magnitude of sustained hydrostatic pressure loading. This 
substantially proves that the ultimate strength of acrylic windows is 
not a function of stress but of strain and that calculations of window 
failure under long-term loading based on stress alone are of little value. 
Cyclic Loading 
The catastrophic failure of EDU windows under cyclic pressure loading 
was found to be very predictable (Table 5). The mode of failure was 
similar to short-term and long-term loadings. The relationship between 
the implosion pressure and number of cycles could be graphically repre- 
sented as a straight line on log log coordinates (Figure 10), and thus 
easy to extrapolate. The windows were found capable of withstanding more 
than 1010 cycles each (7 hours duration at 450 psi pressure) prior to 
requiring replacement due to catastrophic failure. How many cycles they 
will withstand at longer, or shorter than 7 hour cycle loadings is not 
quantitatively known. It is, however, qualitatively known from the NEMO 
experimental program@ that if the duration of an individual fatigue cycle 
on acrylic plastic is less than 7 hours then the fatigue damage to the 
window for each cycle fatigue will be less, and if the duration of a 
cycle is longer, the fatigue damage accomplished by each cycle will be 
greater. But even if the duration of individual cycles was 100 hours, 
it is estimated that it still would take at least 1000 cycles to failure. 
Proof Testing 
All windows were proof tested (Figures 11 and 12) under 50 percent 
overload prior to shipment for installation at EDU. All windows with- 
stood the l-hour long proof test successfully without visual or photo- 
elastic detectable permanent deformation or cracks. 
CONCLUS IONS 
The design, material, and fabrication method chosen for EDU windows 
have been found more than adequate for the service in man-rated hyper- 
baric chambers designed to operate under 450 psi maximum operational 
pressure and ambient temperature not to exceed 120°F. 
10 
