IFMASS10

136

catastrophic failure, but any significant indication of inadequacy for practical use. For

example, the leakage in service may cause the dangerous presence of inflammable or

toxic gases. In this case the wall penetration by the crack in proof test must be treated as

complete failure, even the accident did not occur and the crack behaviour was stable.

Figure 12: Schematics of proof test by fracture mechanics

It is important to understand that the crack size

in the proof test is assumed on the

calculation base, what means that this crack may not exist in reality and must be not

detected during NDI (embedded crack). In general, the determination of

is more simple

than of

, because the condition for proof test performance is very well known and

controlled and, of course, les complex than the conditions in service.

Proof test philosophy is most simple for the brittle materials, where there are no

difficulties connected with the stable crack growth before fracture. In addition, since the

crack tolerance of brittle materials is lower, crack maximum size for fracture is lower.

This allow higher

ratio for given

than for ductile materials, and this allows

longer residual life. It is advantageous to perform the test at the temperature at which the

ductility of the material is lower than in service conditions, although the cost of the test

due to higher probability of fracture can be higher. In ductile materials crack of corres-

ponding size can grow during test without fracture, so that condition after test become

worse and the component can also be less reliable in service than without this test. In

order to prevent this, the inspection after test (and during the test) is requested, and this

can make the all procedure more expensive.

Although

may not be size of real crack, its application does not mean reduction of

the available life. In contrast, by this procedure only the definite period of use without

risk is guaranteed, and the further life extension is after repeated proof test, in order to

again confirm applicability of components for next extended service. Moreover, in this

way the real life capacity (above minimal calculated life) based on random scatter of the

items distribution can be suggested. At the same time this means that the worst case

scenario is temporary and only valid up to the next inspection and repeated proof test.