Comparison study of measured stresses, and stresses from a global analysis model in an offshore structure member

Abstract

The first part of the thesis is related to theory of concern to the thesis and to fatigue life estimation in general. Secondly an investigation of which parameters that introduce uncertainty is investigated. And thirdly a case study related to the comparison of measured stresses by use of strain gauges and transfer functions extracted from the global analysis model of the structures have been compared.

The comparison is made by comparing response amplitudes from strain gauge measurements during a storm and the transfer functions for the relevant directions. This imprinted by very rough estimates and the level of response is not well compared. A higher and lower limit for the response is extracted from the transfer functions. These values differ very much; the higher value is 1.8 times as large as the highest measured stress range. The lower value is 0.16 times the highest measured value.

These values doesn’t make basis for much conclusion with respect to the stress level, except that to some extent the transfer functions seems reasonable. The higher value is found by evaluating transfer function related to sea from 45 degrees heading, and the lower by the one related to head sea. The measured are a product of something in between. The results leads in any case not to any enhanced doubt in the structural model.

The most important lesson learned from that comparison is the response’ sensitivity to wave direction.

Another interesting thing found by comparison is the relation between responses in the elements relative to each other. Response in elements corresponding to the strain gauges is related to each other in a similar way as the response from the measurements do. This implies that the structural behavior is reproduced well by the finite element model.

It would be preferably to calculate a response spectrum from the analysis by use of different direction spreading; this work was started but didn’t reach its goal. More on this in appendix 1.

The main lesson learned by investigating parameters by theory study was related to long term statistics. When comparing inspection results to design expectations long term variations may introduce significant deviations. In [8] it’s stated that the variations in predicted fatigue damage by use of 1-year, 2-year and 4-year deviates considerably with the 29-year average which can be associated with design. The ratio between the smallest and largest predicted damage was. 1.9.

The high sensitivity shown in particular to direction emphasize that the ratio between measured and simulated stresses should be dealt with care, and comparison should not be done without investigating the parameters with respect to uncertainty and sensitivity laboriously.

The damage ratio is related to the power of 3 ref equation 21. And the ratio of damage then will be rapidly very large when differences in stresses.