

MILHDBK470A: Designing and Developing Maintainable Products and Systems 
 

B.3.6  Statistical Maintainability Demonstration Plans
The following subsections contain test methods and criteria for demonstrating the achievement of specified quantitative maintainability requirements. The matrix presented previously in Table BI is repeated as a summarization of the major characteristics of each test method as well as the quantitative requirements which must be specified for each. The data analysis method included with each test method provides the decision criteria for acceptance or rejection of the item being demonstrated.
Each of the test plans contained in this section includes an equation or other directions for determining a minimum sample size of maintenance tasks. Any departure from the minimum sample size requirements can affect the statistical validity of the test procedures. Some of the test plans require a prior estimate of the variance of the distribution of interest for the calculation of sample size. Such prior estimates are typically obtained from data on similar systems provided similarities in maintainability design, skill levels of maintenance personnel, test equipment, manuals and the maintenance environment are considered in the estimation process. Equations for predicting the variance when prior estimates are not available are presented in RADCTR69356 (see footnotes), which can be used, provided the information needed for the prediction is available. To preserve the desired risk values in those cases where the variance is predicted, the 85th 95th upper confidence bound on the predicted or estimated variance should be used. Average values of the variance have ranged from 0.5 to 1.3.
Test Method Matrix
Test Plan 
Test Index 
Assumptions 
Sample Size 
Sample Selection 
Specification Requirements 
1A 
Mean 
Lognormal dist. and prior knowledge of
variance 
See
test method 
Natural occurring failures or
stratification^{1} 
H_{0}, H_{1}, , 
1B 
Mean 
No
dist. assumption, prior knowledge of variance 
See
test method 
Natural occurring failures or stratification 
H_{0}, H_{1}, , 
2 
Critical Percentile 
Lognormal dist., prior knowledge of
variance 
See
test method 
Natural occurring failures or stratification 
H_{0}, H_{1}, , 
3 
Critical Main. Time or Manhours 
None 
See
test method 
Natural occurring failures or stratification 
H_{0}, H_{1}, , 
4 
Median 
A specific Variance Lognormal 
20 
Natural occurring failures or stratification 
ERT 
5^{2} 
Chargeable maint. Downtime/Flight 
None 
See
test method 
Natural occurring failures 
ORR or A, NCMDT/NOF DDT/NOF, , 
6^{3} 
Manhour Rate 
None 
See
test method 
Natural occurring failures 
Manhour Rate MR 
7^{4} 
Manhour Rate 
None 
See
test method 
Natural occurring failures or
statification 
µ_{R }, 
8 
Mean and Percentile 
Lognormal 
See test method 
Natural occurring failures or Simple random
sampling 
Mean, M_{Max} 
Dual Percentile 
None 
Dual Percentile 
9 
Mean
(Corrective Task Time, Prev. Maint. Time, Downtime) 
None 
30 minimum 
Natural occurring failures or stratification 
µ_{c
}, µ_{pm }, µ_{p/c },
M_{Maxc} 
M_{Max} (90 or 95 percentile) 
10 
Median (Corrective task time, Prev. maint. time),
M_{Maxct} (95 percentile),
M_{Maxpm} (95 percentile) 
None 
50 minimum 
Natural occurring failures or
stratification 
_{ct} , _{pm}
M_{Maxct} ,
M_{Maxpm} 
11 
Mean (preventive maint. task
time) M_{Max} (preventive maint. task time at any
percentile 
None 
All possible tasks 
All 
µ_{pm}M_{Maxpm} 
Notes for Table BI:
(1) See section
3.3 for stratification method for sample selection (2) Test method 5
is an indirect method for demonstrating operational ready rate (ORR) or
Availability (A). (3) Test method 6 is intended for use with aeronautical
systems and subsystems. (4) Test method 7 is intended for use with ground
electronic systems where it may be necessary to simulate
faults.
Because of the difficulty in obtaining prior information and estimates of variance, and due to the fact that the mean corrective maintenance time (M_{ct}) and maximum corrective maintenance time (M_{Maxct}) have historically been the maintainability requirement most often cited in a procurement specification, test method 9 has been the most chosen method for maintainability demonstration. Note that this method does not rely on any assumptions regarding the distribution of maintenance times. Despite this fact, there are examples when one or more of the other test methods have been employed, and therefore all methods will be given equal consideration in this handbook.




 
 