OS-C501
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Offshore Standard DNV-<strong>OS</strong>-<strong>C501</strong>, November 2013<br />
Sec.10 Component testing – Page 163<br />
from a log(stress)-log(lifetime) diagram for the anticipated lifetime. If more tests are made the requirements<br />
are given in DNV-<strong>OS</strong>-<strong>C501</strong> Sec.4 [8.8.6].<br />
3.3.5 Stress rupture testing for normal safety class: at least one survival test shall be carried out. The specimen<br />
should not fail during the survival test and it should not show unexpected damage. The requirements to the test<br />
results are:<br />
— tests should be carried out up to three times the maximum design life with realistic mean loads that the<br />
component will experience. If constant load testing is carried out tests should be carried out up to 30 times<br />
the design life to compensate for uncertainty in sequence effects.<br />
— if the anticipated lifetime exceeds 1000 hours testing up to 1000 hours may be sufficient. The load levels<br />
should be chosen such that testing is completed after 10 3 hours. The logarithms of the two test results shall<br />
fall within µ-2σ of the logarithm of the anticipated lifetime, where µ is the mean of the logarithm of the<br />
predicted lifetime and σ is one standard deviation of the logarithm of the predicted lifetime, both interpreted<br />
from a log(stress)-log(lifetime) diagram for the anticipated lifetime. If more tests are made the requirements<br />
are given in DNV-<strong>OS</strong>-<strong>C501</strong> Sec.4 [8.8.6].<br />
3.3.6 For low safety class long term testing is not required.<br />
3.3.7 The sequence of the failure modes in the test shall be the same as predicted in the design. If the sequence<br />
is different or if other failure modes are observed, the design shall be carefully re-evaluated.<br />
3.3.8 The average of the measured number of cycles or time until occurrence of each critical failure shall never<br />
be less than the predicted characteristic lifetime or numbers of cycles. Critical failure modes are failure modes<br />
that are linked to a limit state.<br />
3.3.9 Tests should be carried out with a typical load sequence or with constant load amplitude. If a clearly<br />
defined load sequence exists, load sequence testing should be preferred.<br />
3.3.10 Whether reduced test times compared to the component's life are acceptable should be evaluated based<br />
on the anticipated failure modes and whether extrapolation of the data to longer lifetimes is possible. This will<br />
mainly depend on the confidence and previous knowledge one has about the failure modes that are tested.<br />
3.3.11 In some cases high amplitude fatigue testing may introduce unrealistic failure modes in the structure.<br />
In other cases, the required number of test cycles may lead to unreasonable long test times. In these cases an<br />
individual evaluation of the test conditions should be made that fulfils the requirements of [3.3.2] or [3.3.3] as<br />
closely as possible.<br />
3.3.12 The static strength of the structure after long term exposure shall be taken as the extrapolation of the<br />
long term test data of the fatigue or stress rupture tests.<br />
3.3.13 Higher static strength values after long term exposure may be used if experimental or theoretical<br />
evidence can be provided. The same arguments as given in Sec.4 [3] may be used for matrix and fibre<br />
dominated properties. A procedure to obtain strength data after long term exposure is suggested in Sec.4 [3.4]<br />
and [3.9].<br />
3.3.14 Additional tests may be required if resistance to a failure mode cannot be shown by analysis with<br />
sufficient confidence and if this failure mode is not tested by the tests described above.<br />
3.4 Procedure for updating the predicted resistance of a component<br />
3.4.1 The resistance of the component is R and is assumed to be normally distributed:<br />
R∈N(µ R ,σ R 2 )<br />
where,<br />
µ R = mean value of the resistance of the component (generally unknown).<br />
σ R = standard deviation of the resistance of the component, representing the natural variability in the material<br />
properties and the manufacturing/production process, and here assumed known.<br />
3.4.2 The characteristic value of the resistance is specified as a specific quantile in the distribution of the<br />
resistance, here defined as:<br />
x C =µ R -2σ R<br />
However, because µ R is unknown, the true characteristic value x C is also unknown.<br />
3.4.3 Estimates of µ R and x C prior to testing are sought. One way of obtaining such prior estimates is to carry<br />
out an analysis of the component by means of available analysis models.<br />
The estimate µ RA of µ R is obtained from a single analysis using mean values for the material properties. The<br />
uncertainty in the estimate µ RA should also be assessed, expressed in terms of a standard deviation σ m ’, and<br />
DET NORSKE VERITAS AS