This paper proposes the following processes due to some two principles to get some sauction of early failure for marine equipement.
P1. to divide time after ship in service into a quasi-early failure term and a quasirandom failure one.
P2, to get average failure rate Μ_ó, Μ_í of their terms respectively.
P3. the failure condition is considered to settle down the random failure term when the failure rate M is smaller than Μ_í.
P4, consequently let the term till the random failure term the sanction of the early failure term, then to get the average failure rate Μ_ó, Μ_í from both of the failure reports member and the early failure and random failure terms respectively.
P5, to get the percentage of failure cause to total reports at designing, manufacturing and using stages for the early and random terms respectively.
Some results are as follows (six turbine ships, four diesel ships) from the adoptation them to 1470 failure reports from ten ships put in service for the last seven years;
1) each average months for investigation; (32.7, 37), total reports number; (967, 503), their total average failure rate of a kind of ship group Μ_s; (4. 93, 3.41/ month) .
2) total computed months of early failure; (30, 31), each total failure reports of early/random failure rate term (482/485, 125/378), their average failure Μ_ó; (15.9, 5.09), Μ_í (2.95, 2.26/month) .
3) the percentage of failure cause at manufacturing and useing stage, in the early failure term (78. 1/21.9, 55. 1/44.9), in the random (50.6/49.4, 40.0/60.0 C%J) .
Furthermore the following failure transition modes can be postulated due to the comparation among Μ_s, Μ_ó and Μ_í.
I {Μ_ó, Μ_í<Μ_s} ; the most stable, II {Μ_o>Μ_s>Μ₁} ; stable
III {Μ_ó<Μ_s< Μ_í, } ; unstable, IV {Μ_ó, Μ_í>Μ_s} ; the most unstable
Then these modes can interprete considerablely if the enviroment and level of design, manufacture and treatment for marine equipement are stable or unstable, it is proved that there are some good relation between these modes and the “characteristics” of the early failure.
It is testified that these processes are reasonable to be adopted to ten cargo ships, but should be more improved by many other examples.
It is noted that they can be not only practiced by a computor but also extended to other problems concerning to maintenance and warning alarm problems as well as the early failure of marine equipement.

View full abstract