Você está na página 1de 5

Study on the Fatigue Safety Assessment of Ship Structure Using Risk-Based Approach

Choung-Ho Choung1), Jae-Hong Park1), Tae-Bum Ha1)


1)

R&D Center, Korean Register Daejeon, Republic of Korea

Abstract
In these days risk-based approach is considered as an effective methodology for promoting the safety of ship and the relevant researches and works are actively going on. The SLA (Safety Level Approach) based GBS (Goal-Based new ship construction Standard) in IMO is seen as a representative instance. Developing risk-based method based on the SRA (Structural Reliability Analysis) for ship structural safety assessment is the main purpose of this study. Especially focusing on the fatigue safety assessment for ship structure, several limit state equations and related uncertainties were formulated and assumed. Then, fatigue safety level, which corresponds to the reliability index or probability of failure, was calculated for each limit state equation. As a result, the feasibility of applying risk-based approach to the design and safety assessment of ship structure was examined and demonstrated.

Keywords
Risk-based approach; Structural reliability analysis; Fatigue; Spectral analysis; Safety level; Reliability index; Probability of failure

Introduction
In the maritime industry, there have been lots of activities of introducing risk-based approach into ship design and approval for the sake of further enhanced safety of ship. The most prominent issue at present is the SLA (Safety Level Approach) based GBS (GoalBased new ship construction Standards), which has been initiated and being still discussed by IMO since 2002. The application of risk-based approach to ship is expected to broadly affect the ship design and approval process. However, a specific risk-based procedure, especially for the design and safety assessment of ship structure, has not been fully established and furthermore relevant examples are not sufficient. This study, therefore, was carried out in order to develop a concrete method and detailed example for the ship structural design and safety assessment by using risk-based

approach. When developing a risk-based procedure for the structural safety assessment of ship, the essential point is the computation of quantitative safety level considering functional requirements or failure modes of ship structure. In this regard, the SRA (Structural Reliability Analysis) was chosen as the most appropriate technique. The SRA deals with uncertainties being related to various structural elements in a probabilistic manner and produces the reliability index or the probability of failure as a quantitative safety level of this study, In structure. focusing on the fatigue strength among the several failure modes of ship structure, four kinds of limit state equations for fatigue failure were developed and then the fatigue reliability index for each equation was calculated by using the SRA technique. The spectral analysis (full stochastic method) was used for calculating accumulative fatigue damage ratio based on the Palmgren-Miner rule and S-N curve. As a detailed example of fatigue safety assessment for ship structure, one longitudinal stiffener end connection on side shell plating of a large container carrier was considered. Finally, the calculated fatigue safety levels, that is, the fatigue reliability indices for the abovementioned structure were compared with a specific target safety level proposed by an IMO document (MSC 79/6/15) for ship structure.

Outline of example
In order to calculate accumulative fatigue damage ratio and fatigue reliability index, a specific structural element and its configuration was selected (refer to Fig.1). To be concrete, the subject vessel is a large container carrier (8,600 TEU) and the hot-spot point considered is the toe of web stiffener (150x10 F.B) attached to the flange of side shell longitudinal stiffener, which is located amidship and just below ships draught. This example structure is usually considered as one of the fatigue critical connections in the ship.

Table 1: Conditions for spectral fatigue analysis

Fig. 1: Structural Element and Hot-Spot Point

Item Design life Loading condition Operating condition Service area Stress type S-N curve Corrosion environ. Wave spectrum

Description 20 years full loading condition (only) 85% operating World-wide Hot-spot (extrapolation) 1-slope D curve (U.K Den) Air or Cathodic Pierson-Moskowitz

While designing ship structure, fatigue damage ratio (accumulative) can be calculated by either simplified fatigue analysis method or spectral fatigue analysis method. Spectral fatigue analysis method (full stochastic) was used in this study. For performing spectral fatigue analysis, at first fullship FE analysis should be carried out. Fig. 2 and Fig. 3 show the fullship FE model (coarse mesh) and the detailed FE model (fine mesh) for extracting relevant hot-spot stress.

Limit state equations


For calculating fatigue safety level, i.e. fatigue reliability index, by using the SRA method, four kinds of fatigue limit state equations were invented like Table 2.
Table 2: Limit State Equations

Fig. 2: Fullship FE Model (Container Carrier)

Fig. 3: Detailed FE Model having Fine Mesh

Design conditions for calculating fatigue damage ratio were assumed like the followings;

In Table 2, k denotes the frequency per each wave heading angle and p ij means the occurrence probability of each short-term sea state within the considered wave scatter data. Fundamentally all the limit state equations are based on both the Palmgren-Miner rule and accumulative fatigue damage ratio. LSE-1 has the form of fatigue damage ratio calculation by using simplified analysis, which considers the hotspot stress range of long-term sea state has the Weibull distribution. The mean value of stress range and Weibbull shape parameter at the exceedance probability level of 10-4 was decided respectively from the results of spectral analysis (long-term prediction). LSE-1 means fatigue failure takes place if the accumulative long-term damage ratio calculated is greater than the allowable maximum value, (g 0). LSE-2 has the form of fatigue damage ratio calculation by using spectral analysis, which considers the hot-spot stress range of short-term sea state has the Rayleigh distribution. LSE-2 is more simple and easier to understand its meaning than LSE-1. Firstly the probability of failure for each short-term sea state

failure for each short-term sea state should be computed. For each short-term sea state, if the accumulative damage ratio is greater than the allowable maximum value, (g 0), fatigue failure takes place and corresponding failure probability is calculated. And lastly, the calculated probability of failure for each short-term sea state should be all summed using the wave scatter diagram and then we can find the probability of failure (long-term) throughout the ships design life. It should be noted that the accumulative short-term damage ratio of LSE-2 contains the occurrence probability of each short-term sea state. LSE-3 is almost identical to LSE-2. However, the accumulative short-term damage ratio of LSE-3 doesnt include the occurrence probability of each short-term sea state. The occurrence probability is reflected directly in the probability of failure for each short-term sea state. Thats the difference between LSE-2 and LSE-3. LSE-4 also has the form of fatigue damage ratio calculation from spectral analysis, but looks very different from LSE-2 and LSE-3. According to the procedure of spectral fatigue analysis, the accumulative long-tem fatigue damage ratio is calculated and then eventually whether fatigue failure takes place or not is evaluated. (if the damage ratio obtained is larger than the allowable maximum value, fatigue failure occurs.) In this respect, the concept of LSE-4 is similar to LSE-1. For each limit state equation above, the relationship between short-term calculation and long-term calculation is briefly like the below; l LSE-1 & LSE-4 short-term damage ratio long-term damage ratio long-term P failure l LSE-2 & LSE-3 short-term damage short-term P failure adding up all the short-term P failure long-term P failure

S-N curve).
Table 3: Random variables for LSE-1

Fig. 4: Long-term Prediction of Stress Range

In case of LSE-2, LSE-3 and LSE-4, four variables were used in the equations and their characteristics are like Table 4. Two variables are probabilistic and the others are deterministic. Five variables in LSE-1, which are left out from LSE-2 to LSE-4, can be directly determined by the momentum values calculated from spectral analysis results. Thats why much less variables were used in LSE-2 to LSE-4, compared to LSE-1.
Table 4: Random variables for LSE-2 to LSE-4

Random variables
For calculating the probability of failure, the characteristics of variables used in the limit state equations, which may be either deterministic or probabilistic, should be defined. Exactly defining the probabilistic features of variables is not simple work and might need complicated and multidisciplinary investigation. For this example calculation, considering some information already discussed and presented before, the characteristics of variables were assumed and used with discretion. In case of LSE-1, nine variables were used in the equation and their characteristics are like Table 3. Three variables are probabilistic and the others are deterministic. Both Weibull shape parameter (h) and stress range (0) at the exceedance probability level of 10-4 were found from the result of long-term prediction (refer to Fig. 4). It should be noted that S-N curve parameters in Table 3 are without two standard deviation shift (mean S-N curve).

It should be also noted that S-N curve parameters in Table 4 are without two standard deviation shift (mean S-N curve).

Deterministic fatigue safety level (fatigue life)


As a result of spectral fatigue analysis, accumulative fatigue damage ratio for ships design life and corresponding fatigue life can be obtained. This method is basically deterministic approach and thus cannot consider various uncertainties of related variables. Fatigue life computation by using spectral analysis is usually applied to the fatigue safety evaluation of ship and offshore structure in these days. The calculated fatigue life for the concerned structure is approximately 23.8 years and larger than the design life initially defined.

l Accumulative fatigue damage ratio : 0.84 l Fatigue life : 23.8 (years) Consequently, it can be said that the structure has sufficient fatigue safety level from a deterministic point of view. It should be noted that for computing the fatigue life above, design S-N curve, which was modified to contain two standard deviation shift, was applied. For reference, the calculated fatigue damage ratio for each short-term sea state is shown in Table 5.
Table 5: Accumulative short-term fatigue damage ratio

Table 8: Specific expression of LSE-3

Table 9: Specific expression of LSE-4

Table 10: Probability of fatigue failure from LSE-3

Probabilistic fatigue safety level (reliability index)


Considering the characteristics of variables defined in Table 3 and Table 4, the limit state equations in Table 2 were put in a definite expression respectively such as Table 6 to Table 9. And, in addition, the fatigue failure probability for each short-term sea state, which was from LSE-3, is shown in Table 10, for reference.
Table 6: Specific expression of LSE-1

Table 7: Specific expression of LSE-2

Each fatigue reliability index for the limit state equations above was computed by using the AFOSM (Advanced First-Order Second Moment), which is known as one of the very useful techniques for the SRA. Fatigue reliability index corresponds to probabilistic fatigue safety level. The calculated indices are summarized in Table 11. It should be noted that, for computing the fatigue reliability indices, mean S-N curve, which has no modification for two standard deviation shift, was applied.
Table 11: Fatigue reliability indices calculated

As for LSE-2, the probability of failure is almost zero and considered as a meaningless result. So, it can be said that the definition of fatigue limit state equation such as LSE-2 is not correct. In case of LSE-3, the probability of failure is about 10% for the design life of 20 years. This probability is considered excessive and the fatigue safety estimated by LSE-3 is much conservative. The failure probability results of LSE-1 and LSE-4 are

considered relatively reasonable, based on the past experience of ship structure design. The probability of failure for LSE-4 is about 0.1% for the design life of 20 years. It physically means that when there are 1,000 structural elements having totally equivalent configuration, one of them might have a crack by fatigue failure right after 20 years. It is noticed that the estimated failure probability by deterministic approach with two standard deviation shift is approximately 0.05%. For the sake of assessment of fatigue safety and decision-making process for ship structure design, it is required not only to calculate the present safety level, but also to define the target safety level for comparison. (Of course, both are to be quantitative.) In this study, taking into account there is no rigorous target safety level internationally agreed for ship structure, one of several instances already presented in public was quoted like Table 12.
Table 12: Example of target safety levels (MSC 79/6/15)

fatigue safety level of ship structure and also to be utilized for the risk-based assessment of fatigue strength of ship structure in a probabilistic way, not deterministic.

Future work
Actually precise definition of random variables is very critical to the accuracy of calculated safety level and/or reliability index. Therefore, further study of a detailed and thorough investigation for random variables related to fatigue safety should be carried out in future.

References
KR (2009), Rules and Guidance for the Classification of Steel Ships, Part 3, Hull Structures, Annex 3-2, Guidance for the Direct Strength Assessment, Korean Register, Korea. KR (2009), Rules and Guidance for the Classification of Steel Ships, Part 3, Hull Structures, Annex 3-3, Guidance for the Fatigue Strength Assessment of Ship Structures, Korean Register, Korea. Cornell, C. A. (1967), Bounds on the reliability of structural systems, Journal of structural division, ASCE, vol.93, no.ST1, February. Freudenthal, A. M. (1956), Safety and the probability of structural failure, Transactions of ASCE, vol.121, pp.1337-1397. Hasofer, A. M. and Lind, L. C. (1974), Exact and invariant second moment code format, Journal of the engineering mechanics division, ASCE, vol.100. Hohenbichler, M. and Rackwitz, R. (1981), Nonnormal dependent vectors in structural safety, Journal of the engineering mechanics division, ASCE, vol.107, no.EM6, December, pp.1227-1238. Ichikawa, M. (1993), A meaning of the overlapped area under probability density curves of stress and strength, Reliability Engineering and System Safety, vol.41, pp.203-204. Rackwitz, R. and Fiessler, B. (1976), Note on discrete safety checking when using nonnormal stochastic models for basic variables, Loads Project working session, MIT, Cambridge, Mass., June. Rackwtiz, R. and Fiessler, B. (1978), Structural reliability under combined random load sequences, Computers and structures, vol.9, pp.489-494. Rosenblatt, M. (1952), Remarks on a multivariate transformation, The annals of mathematical statistics, vol.23, no.3, pp.470-472. Shinozuka, M. (1983), Basic analysis of structural safety, Journal of structural division, ASCE, vol.109, ST3, March, pp.721-740.

Fatigue failure on the end connection of longitudinal stiffener on side shell plating may be considered as both Ductile failure with no reserve capacity and Not serious, which are described in Table 12. Thus, target probability of failure, P failure-target becomes 2x10-3 (=10-4 x 20 years). In case of LSE-4, the calculated probability of failure, P failure-actual is about 8.76x10-4 and smaller than the target value. Consequently, it can be said that the concerned structure has sufficient fatigue safety level from a probabilistic point of view when using LSE-4. However, if LSE-1 is applied, P failure-actual becomes 9.21x10-3 and larger than the target value. Then, it can be said that the concerned structure has insufficient fatigue safety and additional design amendment is required when using LSE-1. Accordingly, it should be noted that the probabilistic approach and the deterministic approach may result in different evaluation and different decision-making process of ship structure design.

Conclusion
In this study, for the quantitative calculation of fatigue safety level of ship structure by using structural reliability analysis, several specific limit state equations were suggested and their applicability was examined as well. Considering the SLA based IMO GBS framework to be developed in future, this study is expected to provide practically useful example and method for calculating fatigue safety level of ship structure and also

Você também pode gostar