Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 Fatigue problems in ship structure J. Kozak Faculty of Ocean Engineering and Ship Technology, Technical Univ. of Gdansk, 80-952 Gdansk, Narutomcza 11/12, Poland E-mail: kozak@pg.gda.pl Abstract Fatigue strength of ship structure is one of the components of reliability of ship as a mean of transport. Majority of fatigue strength assessment proposals given by classification societies are based on use of characteristics of typical joints or designs of hull construction given if form the Wohler curves. More accuracy calculation will be done if more adequate Wohler curve to calculate the problem will be applied. To gather wide spectrum of fatigue properties of typical ship hull nodal points fatigue tests should be performed and their results have to be systematised. Paper show an exemplary of systematic fatigue tests result as well as way of their systematisation. Introduction During our life rapid growth of new types of ships and offshore structures is observed and completely new and narrow specialised designs have birth. Due to the increasing amount of different cargoes potentially dangerous for environment the construction of ships and its reliability have to be permanently improved. Even the new potentially safer construction with large or innovative designs like the new generation of double hull tankers need of the knowledge of their behaviour under complex service load and often in extremal weather or sea conditions. One of the potential modes of the destruction of ship structure is a fatigue as the effect of interaction of such factors like load, material, design, manufacturing and environment'. Such structure like ship hull contains great number of welds, weld crossings, imperfections of manufacturing or assembling etc, which lead to geometrical, structural or technological notches creating favourable conditions for fatigue crack Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 542 Marine Technology initiation and its further growth. In general, ship hull is a thin skin structure supported by system of longitudinal and transverse stiffeners. Additionally safety and ecology requirements have modified classical designs of ships adding double skin plating. As the consequence of those is fact that in general ship hull almost in majority consists of large number of plate crossings, weldings, stiffener-stiffener or plate-stiffener intersections, cut-outs and wholes. All of such components are 'calculated and manufactured difficult' and their load capacity is connected with quality of their manufacturing. Moreover depending on significance of analysed detail for fatigue strength of ship hull, the range of fatigue processes and as a consequence the range of the problems of fatigue in ship structure can be divided into three groups: local, regional and global \ Local range of fatigue strength problems consists of such class of defects like welds or cruciform joints whereas global problem of fatigue can be identified for instance with problem of deck cracking in the wide region of the bow edge of the superstructure. In general, total reliability of structure is an effect of combination of small, medium and large scale processes of fatigue. Looking at simplified cross section for instance of Ro-Ro ship one can extract several places in which the problems of fatigue strength can occur. There are such points like bilge corner, double skin-twindeck joint, cross girder connection or girder stiffener intersection as it is shown in Fig.l. Proper calculation of fatigue strength of such complex structures requires - as initial data - the knowledge of at least two parameters : fatigue properties of analysed body and load conditions as a sources of failure. Both groups of factors are very difficult to define. First one - because of fact that fatigue life properties are strongly depended on quality of manufacturing as well as state of load. The second one - because of fact that all mentioned places are subjected to mixed and complicated state of load and it is difficult in practical design of ship structural detail to define the proper nominal stress level. Furthermore fact that process of calculation should be supported by proper definition, not only the level, but of state of the load also and by selection of adequate load representation what creates the next complication. Moreover fatigue properties of particular structure design are connected with way of loading and supporting scheme. As the fatigue analysis is not a rigorous science it is difficult to compute the exact life of a structure. The reason is that some data involved are subjected to some degrees of uncertainty *. One of group of such data is load that is combined by set of such components like still water loads, wave bending loads, dynamic loads and thermal effect loads. Mentioned above manufacturing imperfections are another important groups of non-full determined factors too. Additionally the majority of the ship structural components are more complex than the test specimens used for determination of the strength patterns, both in geometry as well as in applied load and a relationship between the S-N data stress and the calculated stress can not be easily obtained Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 Marine Technology 543 Figure 1. Middle cross section of small Ro-Ro ship Fatigue assessment of the complex structures During some last years several proposals for the assessment of the fatigue of the ship hull structure have been presented. The general scheme proposed is to split the structure into elementary substructures and to search for properties of each selected potentially 'weak chain' of construction. A fatigue tests and theoretical approach are two major paths for the prediction of the fatigue behaviour of the ship hull structure. Classification societies developed the fatigue assessment procedures, which - beside different approaches to details - in general are supported on utilising the design fatigue curves because of its simplicity for practical application *. Such approach assumed that ship hull structure is divided into simple, elementary details for which the fatigue behaviour is known and given in form of fatigue (Wohler) curves. Of course methodology that allows performing such separation process as well as to define load conditions or determine criteria for damage is a key point of all developed assessment procedures and is different in particular approach. To adjust Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 544 Marine Technology way of fatigue failure as close as possible to reality the fatigue characteristics of the welded joints have been divided into several classes - each of them with a corresponding design S-N curve obtained from experiments. Independent on applied methodology the common platform of all approaches is utilisation of the design curves. The exemplary, basic design curves are shown in Fig.2/ 300 A<7 [N/mrrf] 100 :B :c 50 \ D E F F2 30 G -w 20 10' 10' 10' 10" N[cycles] Figure 2. UK Dept of Energy Design Curves Main and common source for fatigue characteristics of design details which are transformed into design curves shape are laboratory carried fatigue tests because only such way gives the basis for structural analysis of the properties of the tested node. Mentioned above simplification of welded joints and way of cracking lead to simplicity of the calculated model. So, the uncertainty margin is relatively wide because of differences between the analysed structure and typical representing particular class. There is no possible also to take into account the effect of the Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 Marine Technology 545 sequence of load in a case of the various amplitude loading. The knowledge of behaviour of crack during its propagation stage as a very important indicator for designer as well as explorer of construction can make the calculation process more precise. To find the answer for more complex and detailed problems of cracking the another methods for assessment of the fatigue behaviour of structure should be used. One of them is a theoretical approach. Packet of tools for this methodology gives linear fracture mechanics ^. In general the formulae for crack propagation velocity is defined basing on stress intensity ratio. The integration of such formulae gives fatigue propagation data \ The obstacle of easy using of this approach is proper obtaining of the strict solution for stress intensity factor formula but uncertainties in results can be overcome by using a numerical calculation. However at a present stage of development this methodology is too sophisticated for design and exploitation practice. Laboratory tests of real scale models of bilge corner of Ro-Ro ship As it was mentioned above it is possible in hull structure to identify places which are important from the fatigue strength point of view. Some of such places are presented in Fig.l and majority of them was investigated during last several years in Technology Laboratory of Shipbuilding Faculty of Technical University of Gdansk. Those are marked with a circle in Fig. 1. Among others investigations were made for fatigue properties for family of bilge corner *. Some variants of connection of inert skin and inert bottom plating were investigated ^. Model design, load and support conditions are shown in Fig. 3 whereas in Fig.4 places and trajectories of developing fatigue cracks during the tests are shown. To make possible the presentation and comparison of fatigue test results in qualified shape the test data has to be homogenised. Problem arises when one tries build the Wohler curve for whole family and then makes comparison with tests results of another geometry, because stress as determinant of one axis on diagram should be normalised because different models had been tested with different load level. Decision what stress level would be used, as a reference should be made. The nominal theoretical stress level caused by bending was applied. It created next obstacle: how reduce effect of width of shell plating for different geometry because this factor has got influence both on real stress distribution as well as on theoretical bending stress level. So it is important to keep in mind when using the design S-N curve in the fatigue assessment process the calculated stress should correspond to the stresses used in building the curve. From that point of view it is important to know which type of adequate stresses is to be used for fatigue assessment: nominal, hotspot or notch stress. Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 546 Marine Technology Loading & supporting scheme r " \ / 9 X/a 75*6 0 0 J Detail "a" : model 21 e Detail "a" : models 22-20 Detail "a" : model 27 Detail "a": model 20 Figure 3 Details of models of bilge corner of Ro-Ro ship. Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 Marine Technology Model 21 Model 22 Model 23 Model 24 Model 25 Model 26 Model 27 Model Model 29 Figure 4. 547 Way of cracking of models of bilge corner of Ro-Ro ship. Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 548 Marine Technology To prove above Fig. 5 presents fatigue tests results of whole family of tested models compared with mentioned design curve class F2 suggested by majority of proposals for welded joints. One can see that for some presented designs fatigue tests results of really tested construction are located below of field limited by lines of +/- two standard deviations from Wohler line F2 class. It means that real fatigue strength is less than would be obtained from calculation process. Once more the significance of proper selection of the design curve for calculation and as basis for such selection necessity of building of catalogue of such curves should be pointed out 1000 Models 20 100 JQ CO 10 F2-2s M22 M25 M28 1E+4 Figure 5. 1E4-5 1E+6 F2+2S M23 M26 M29 1E4-7 1E4-8 N[cycles] Fatigue tests results of models of bilge corner of Ro-Ro ship. Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 Marine Technology 549 Conclusions The fatigue assessment of ship hull structures is still very complicated problem and there is still a lot of gaps in methodologies in spite of wide development of procedures and analytical tools, In majority fatigue assessment procedures suggested by classification societies are supported on Safe-Life approach on design curves which represents modified Wohler curves, For wide spectrum of geometry and load condition in ship hull structure it is difficult to create a fatigue assessment procedure which utilise only one, common design curve, Linear fracture mechanics is alternatively tool for fatigue assessment in ship hull structure but at present stage of development it is too sophisticated for design and exploitation practice, Laboratoiy carried out fatigue tests of huge scale of details of ship hull structure are still important source for data both for creation or verification of new design procedures as well as simply for solution of problems which can not be solved analytically. References 1. Dzieduszycki K., Gorski Z., Kubera S., Rosochowicz K., Szczepariski J., Optymalizacja wytrzymalosci zm^czeniowej w^zlow konstrukcji okretowych ze stopow aluminium, prace IOPG Gdansk 1975. 2. Rosochowicz K., P?kni^cia zm^czeniowe w konstrukcjach kadlubow statkow i innych obiektow oceanotechnicznych, metody przewidywania i napraw, Marine Technology Transactions, Vol. 7, pp.231-254, 1996 3. Radaj D., Review of fatigue strength assessment of nonwelded and welded structures based on local parameters, Int. J. of Fatigue, Vol 18, No. 3, pp. 153-170, 1996.Middle cross section of small Ro-Ro ship. 4. Rules of Classification Societies. 5. I ACS, Report on the Development of a Unified Procedure for Fatigue Design of Ship Structures, December 1996 Transactions on the Built Environment vol 42, © 1999 WIT Press, www.witpress.com, ISSN 1743-3509 550 Marine Technology 6. Fatigue Strength analysis for Mobile Offshore Units, Det Norske Veritas, Classification Notes, Note No 30.2, 1984 7. Kozak J.:"On calculation of the stress intensity factor for numerical assessment of the fatigue crack propagation", Computational Mechanic Publications, 1997. 8. Borzgcki T., Dzieduszycki K.,Kozak J.,Kubera S.,Woloszyn A.: "Wyniki badari eksperymentalnych w^zla potyczenia dna podwqjnego z burt^ wewn^trzn^ na statku do przewozu kontenerow. Oprac. IO PG,1980. 9. "Sprawozdanie z badari zm^czeniowych modeli w^zlow konstrukcyjnych pol^czenia burty podwqjnej z mi^dzypokladem statku typu Ro-Ro"., oprac IO PG nr 1171/1979.