I Downloaded from https://www.everyspec.com MIL-STD-81OD 19 July 1983 APPENDIX 516.3A ., DEFINING TEST SHOCK FOR NORMAL OR LOGNORMAL DISTRIBUTION OF FIELD DATA For each 1/6th-octave (or more closely spaced) frequency, the test shock spectrum value G is given by G.m+ks (1) where ‘Ire”is the mean and I’snis the standard deviation of the sample values. .; When the n sample values gi are aseumed on the basis of experience with similar . data to be normally distributed Ql=&~gi and 92 = ~ ~ (gi-m)2 (2} i=1 i=, ~ :
and when the sample values are aasumed to be lognormally distributed m=*jllOggi and 92 = ~- ; (log gi-m)2 i=l (3) The value of k in equation (1) is taken from table 1 as a function of the number of sample values n to yield a G value equal to the 95th percentile at least 50 percent of the time. The.use of this procedurebecomes highly suspect for less than five’data samples. Any such use should be undertaken in association with the approach recommended in I-3.3c when insufficient data are available for statistical analysis. I The above process is repeated for each frequency of the shock response spectrum.
The resulting G valuee may be enveloped in a manner deemed appropriate by the responsible test organization. TABLE A-I. YaMM2f& ~j No of ‘ No of Test Runs k Test Runs k. 5 1.78 12 1.69 6 1.75 15 1.68 -1 1.73 ; 20 1.67 8 1.72 30 1.66 9 1.71 100 1.65 10 1.70 ,. ~ Owen, D.B. Eectors fOr One S@ed TOlerance L -. imits ad for Variablea Sandia Corporation Monograph No. SCR-607, March 1963. MITHOD 516.3 APPENDIX A 19 July 1983 516.3-26