Jurišić, Paul (2012) Utjecaj starenja na graničnu čvrstoću brodskog trupa. = Ultimate hull girder strength of an aging ship. Doctoral thesis , Sveučilište u Zagrebu, Fakultet strojarstva i brodogradnje, UNSPECIFIED. Mentor: Parunov, Joško.

Text
06_12_2012_paul_jurisic_disertacija.pdf Jezik dokumenta:Croatian Download (4MB)  Preview 
Abstract (Croatian)
U radu je provedena analiza korozijskih istrošenja tankera s jednostrukom oplatom na temelju izmjera debljina elemenata trupa iz baze Hrvatskog registra brodova (HRB) s periodičnih pregleda brodova u službi nakon 10, 15 i 20 godina. Matematički model koji opisuje vremensku propagaciju korozije kalibriran je na temelju mjerenja nakon 10 i 15 godina. Predviđanje korozijskih istrošenja na osnovi tako kalibriranog matematičkog modela vrlo dobro se slaže s izmjerenim korozijskim istrošenjima nakon 20 godina. Istraživanje je provedeno kako za globalni gubitak momenta otpora glavnog rebra, tako i za lokalna istrošenja palubne konstrukcije tankera. Definiran je model jednoosno opterećene ploče i ukrepljenog panela primjenom nelinearne analize metodom konačnih elemenata (\NMKE). Model \NMKE je provjeren usporedbom s dvije verificirane metode za proračun kolapsa ploča i panela: metodom iz Usuglašenih pravila za projektiranje tankera, te programom PULS klasifikacijskog društva Det \Norske Veritas. \Na temelju provedene komparativne analize zaključeno je da model \NMKE daje vjerodostojne rezultate pa se primjena modela mogla uzeti u obzir kod daljnjih analiza čvrstoće panela palubne konstrukcije oslabljene djelovanjem nejednolikih korozijskih istrošenja i zamornih pukotina. Ovakvi uvjeti, koji odgovaraju realnim uvjetima istrošenja brodskih konstrukcija, mogu se simulirati pomoću \NMKE, dok su preostale metode koje propisuju pravila klasifikacijskih društava pogodne samo za jednolika korozijska istrošenja. Analizirane su dvije metode proračuna granične čvrstoće: metoda jednog koraka uz pomoć programa PULS i metoda progresivnog kolapsa korištenjem programa MARS, klasifikacijskog društva Bureau Veritas. Pokazano je da metoda jednog koraka daje do 2% niže rezultate proračuna granične čvrstoće u odnosu na metodu progresivnog kolapsa. Predložen je način analize utjecaja nejednolikih korozijskih istrošenja i zamornih pukotina na graničnu čvrstoću trupa inkorporiranjem rezultata kolapsa panela dobivenih \NMKE. U završnom poglavlju rada predložen je niz praktičnih postupaka kao pomoć u proceduri inspekcije i održavanja brodova u službi. Predložene su korozijske mape kojima se vizualno može procijeniti kada će konstrukcijski element trebati zamjenu zbog prekomjernih korozijskih istrošenja. Predložene su i metode planiranja inspekcijskih pregleda koje kombiniraju degradirajuće učinke napredovanja korozije kroz godine službe s fizikalno utemeljenim proračunskim postupcima određivanja nosivosti oslabljene konstrukcije trupa. \Na taj se način međusobno prožimaju metode razvijene u prethodnim poglavljima rada.
Abstract
"Motivation for this thesis lies in the effort and strategic decision of the Croatian Register of Shipping (CRS) for improving knowledge related to the safety of maritime transport. The issue of ship structural assessment at the end of design service life is one of the most complex and important problems faced by the Classification Societies. The scientifically based knowledge of structural behaviour of aged ship may reduce the risk of wrong decisions. Such risk is always present, despite the existing regulations of the International Maritime Organization (IMO). One of the most important questions imposed to the Classification Societies is to define whether the ship at the end of design service life should be repaired or to be sent to scrap and with that will be taken out of service. Also, another critical decision has to be taken, whether fatigue cracks and nonuniform corrosion degradation should be repaired immediately or it is acceptable to postpone repairs until the next ship periodic survey. This is a timeconstrained decision because of the short dry docking period. Both of questions are related to large economic cost from one side and shipping safety and reputation of the Classification Societies from the other side. Development of a new approach dealing with the ultimate hull girder strength analysis of aging ships, adapted for the use in engineering practice, will contribute to the rational decision for maintenance planning, inspection and repair of aging ship. It is expected that the approach, developed in this thesis, will contribute to a better positioning of the CRS within the International Association of Classification Societies (IACS). It is also expected to improve the interaction between the Classification Societies and shipowners. 1. INTRODUCTION The corrosion makes ships prone to structural degradation and the possibility of accidents increases with the aging of ships. Statistical analyses reveal that corrosion degradation is the number one cause for structural casualties in aged ships. The consequences of corrosion wastage can be very serious leading to crack initiation and propagation to such extent to cause the loss of ship. v iii Hull girder section modulus (HGSM) is a fundamental descriptor of the ship longitudinal strength and can be used to measure how the ship hull deteriorates over time because of corrosion. Traditional strength analysis uses simplified deterministic approaches to account for this time dependent deterioration process. Common practice is that nominal corrosion additions are to be predefined. Thus, the Classification Societies Rules, including the newly developed Common Structural Rules (CSR) for Doublehull Oil Tankers [1], assume constant loss of HGSM during the whole service lifetime of a ship. Ship structural assessment, in the design phase, is performed assuming such ""net"" HGSM. Although ""the Rules"" approach is practical, it is obviously conservative one, as HGSM loss is actually a timedependent nonlinear function. Nowadays, there is a clear trend of adopting nonlinear models in order to predict the longterm corrosion wastage degradation and the associated HGSM reduction as a function of time. Developing a direct approach accounting for the real corrosion degradation will be a very important and useful tool for the Classification Societies and ship owners in order to predict the longterm hull degradation and to decide whether the renewal of hull structures are necessary and when would be the optimal time for repairs. The time for repair of aged ships is empirically determined by the Classification Societies, based on existing criteria of damage allowance. One of the most important criteria is the socalled ""10% rule"", which represents tolerated 10% loss from initial asbuilt HGSM. Recently, based on numerous studies and public pressure as well, the approach for longitudinal strength assessment through HGSM of newbuilt ships started to be complemented based on the hull girder ultimate strength (HGUS) assessment, as a physically justified criterion for entire capacity of the ship hull. The logical continuation of this approach in further research is that for the old ships HGUS is used as a criterion of structural adequacy. The results of such approach were used in this thesis. Although the longitudinal strength of the hull structure is in focus as the most important component of the ship strength, the results and findings presented in this thesis can be used in the calculations of other components of strength, primarily in the analyses of the primary transversal structure using the finite elements method (FEM). 2. ASSESSMENT OF HGSM LOSS AND THE LOCAL CORROSION WASTAGE One of the principal studies in this thesis is to analyse the practical applicability of the currently used longterm nonlinear corrosion wastage models and their predictions. The corrosion degradation of three tanker ships after 20 years in service have been analysed based on the real thickness measurements at 10 and 15 years in service. The actual measurements i x after 20 years' service life were then compared to the theoretical model predictions and statistical measures of uncertainty were determined. The assessment of HGSM reduction was performed, followed by the assessment of local corrosion deterioration of main deck plates and longitudinals. For the local corrosion deterioration analysis, total number of 6567 measurements was used. The approach applied here is different comparing to other similar studies, as individual ships are considered, while in other similar studies whole fleets of ships are analysed together. The main goal of the present approach is to fit nonlinear corrosion deterioration models to measured corrosion thickness for 10 and 15 years of service life, and then to assess how well the corrosion deterioration after 20 years is predicted. First, this thesis analysis the corrosion wastage of three oil tankers with singlehull structure built in the eighties. The whole cargo area is made of mild steel for the two sister ships, while the third one has high tensile steel in the bottom and deck areas and mild steel in the neutral axis area including the longitudinal bulkheads. Central tanks along the cargo hold areas are cargo oil tanks, while wing tanks are used for ballast or cargo. The main characteristics of the three singlehull tankers are shown in Table 2.2, and the midship section of the first singlehull oil tanker is shown in Figure 2.11. HGSMs in cargo hold area were calculated using MARS, the program for 2D sectional analysis developed by the classification society Bureau Veritas (BV). The HGSM reduction was defined based on the following procedure. First, the asbuilt HGSM is calculated. Then, thickness of the structural elements (plates and longitudinals) contributing to the longitudinal strength is modified according to the results of thickness measurements from the Croatian Register of Shipping (CRS) file. Gauging records were obtained during the periodic drydockings and closeup surveys of ships in service after 10, 15 and 20 years. Corrosion deterioration models of the ships were determined for the transverse sections considering the central tanks as cargo oil tanks and the wing tanks as ballast tanks. The aging effect was measured by the HGSM loss R(t), which represents the ratio of asgauged HGSM and asbuilt HGSM, as given by eq. (2.7). Results for the measured R(t) for the three ships after 10, 15 and 20 years are presented in Table 2.3. HGSM loss is determined as a function of time taking into account the coating lifetime. The equation (2.8) is used to define the HGSM loss as a function of time has been proposed by Wang et al. [9]. The governing parameters of Eq. (2.8), C and index I are defined based on the regression analysis using a data set of real measurements of corrosion x depth that leads to the HGSM loss. The coating life, t0, which represents the time when HGSM starts to degrade from the asbuilt condition is assumed. In order to analyse the accuracy of the HGSM prediction model, Eq. (2.8) is fitted through the calculated points of the HGSM loss at 10 and 15 years. The governing parameters of Eq. (2.8) are presented in Table 2.5, while the corresponding diagrams are shown in Figure 2.14. It can be seen from Figure 2.14 that the HGSM loss at the 20th year for ship no. 1 is significantly underestimated by the model prediction. The prediction of the HGSM loss at the 20th year for ship no. 2 is satisfactory. The HGSM reduction is almost linear from the time of the coating breakdown at 6th year. The results for ship no. 3 indicate that the coating lifetime is the longest, resulting in 8 years, but after the breakdown of the coating, the corrosion deterioration is very fast. The prediction of the HGSM reduction for that ship overestimates the actual HGSM loss at 20th year. It should be mentioned that there is a large uncertainty associated to the HGSM loss as the corrosion degradation is different for each transverse section of the ship hull. It is assumed that the measured corrosion depths of the sections are representative (average) of the corrosion losses. Additional analyses are required to reduce the uncertainty of the longterm predictions of corrosion reduction of HGSM. The local corrosion wastage analysis is performed separately for plates and longitudinals of the main deck in ballast and cargo tanks. For the local corrosion wastage, equation (2.9) is proposed according to Gua et al. 2008 [10]. In equation (2.9) [10] C(t) is the corrosion wastage at age t; t0 is the year when thickness of the plates starts to deviate from the asbuilt condition; B and C are constants that can be determined according to the measurement data. The total number of measurements of deck plates and longitudinals (stiffeners) used in the thesis are 6567. 2135 measurements were used for ship no. 1, 2079 for ship no. 2, and 2353 for ship no. 3 respectively. Similarly to the prediction model of HGSM loss, curves are fitted through the real measurements of corrosion depth for deck plates and longitudinals at 10th and 15th years. Since there are three unknown variables in equation (2.9) (t0, B and C), and only two calibration points (at 10 and 15 years) are available, t0 has to be assumed. Eq. (2.9) is fitted to the mean value of corrosion depth and also to the value of the 95% confidence level of the corrosion depth, which is related to socalled the extreme corrosion depth. To estimate the extreme corrosion depth, the probability density function of corrosion depth is fitted to the each data set. The Weibull probability density function is x i assumed to be the most adequate for representing the corrosion depth at any time during the ship service life (Parunov et al. 2007 [13], Gua et al. 2008 [10]). Such satisfactory fitting of the Weibull probability density function is confirmed also in the present study, as demonstrated in Figure 2.15 and Figure 2.16, for deck plates in cargo tanks at year 15th and 20th for ship no. 2. The mean value and the extreme corrosion depth of the main deck plates in cargo tanks are presented in Figures 2.17 and 2.18 respectively. It can be seen from Figure 2.17 that the newly developed corrosion model overpredicts the mean value of corrosion depth of ship no. 2 while it underestimates the measured corrosion depth for ship 1 and 3. Coating life is estimated to be 6.5 years. It is evident, as can be seen from Figure 2.18, that the extreme corrosion depth after 20th year is well estimated for ship no. 3. For ship 1 and 2, the corrosion depth after 20th year is underestimated and overestimated respectively. The corrosion depth trend is almost linear for ship 2 and 3, while the coating life is estimated to 5 years for all three ships. It is interesting to notice from Figures 2.17 and 2.18 that for ship no. 2, the corrosion depth after 20th year is only slightly higher compared to the corrosion depth after 15th year. For ship no. 1, in contrary, the difference between the corrosion depth at 20th and 15th years is unexpectedly large, indicating that in this case the concave curve is more appropriate for corrosion process modelling, as it is shown in Figure 2.7 for type A curve (Paik [18]). The mean and extreme values of corrosion depth of main deck longitudinals in cargo tanks are presented in Figures 2.19 and 2.20 respectively. All three estimations overestimate the real corrosion depth measurements after 20th year. That is true for both mean and extreme corrosion depth values. It has been noticed that for deck plates, the corrosion rate is reduced between years 15 and 20 for ship no. 2, which is also confirmed for deck longitudinals. Based on the analysis just presented, one may conclude that the corrosion depth of deck longitudinals is larger comparing to deck plates. Thus, for a 20 years' service life, the extreme value of corrosion depth of deck longitudinals is between 3 and 4 mm, while for deck plates it is between 2 and 3 mm. Coating life of deck longitudinals is estimated to be 8 and 5 years for the mean and extreme value of corrosion depth respectively. The mean and extreme values of the corrosion depth for main deck plates in ballast tanks are presented in Figures 2.21 and 2.22 respectively. For ship no. 1, similar to cargo tank deck plates, the difference between the corrosion depth at 15th and 20th year is quite large, indicating that the type of concave curve should be used for the corrosion model. The tendency of the reduced corrosion rate for the period between 15th and 20th years, which has x ii been already noticed for the cargo tanks, is evident also for ballast tanks of ship no. 2. Consequently, the corrosion depth after 20 years is systematically overestimated by theoretical prediction for that vessel. For ship no. 3, predicted corrosion depth agrees well with the real measurements for both mean and extreme values of corrosion depth. By comparing Figures 2.21 and 2.22 with Figures 2.17 and 2.18, it appears that the corrosion depth is higher for deck plates in cargo tanks than in ballast tanks. The coating life is estimated to be the same as for the deck plates in cargo tanks, i.e. 6.5 and 5 years for the mean and extreme values of the corrosion depth respectively. The mean and extreme values of the corrosion depth of main deck longitudinals in ballast tanks are presented in Figures 2.22 and 2.23 respectively. The estimation for the mean value slightly overestimates the real measurements of corrosion depth after 20th year. For the extreme value prediction, the developed model overestimates the real measurements for ships 1 and 2, while it underestimates the measured corrosion depth for ship no. 3. Generally, measured corrosion depth for deck longitudinals in ballast tanks is approximately the same as for deck plating in ballast tanks. The coating life for deck longitudinals in ballast tanks reads 5 years for the extreme corrosion depth, while between 6.5 and 8 years for the mean corrosion depth wastage. In order to quantify uncertainty of the new developed corrosion degradation model used in the present study, the predicted and measured mean and extreme corrosion depth values (presented in Tables 2.72.9) were subjected to further statistical analysis. The average ratios of the measured and predicted corrosion depth read 1.01 and 0.96 for the mean end extreme corrosion depth respectively, while the corresponding standard deviations read 0.30 and 0.24. This indicates that the new developed model has almost no bias, but uncertainty of predictions is rather large. If only cargo tanks are considered, then the average ratios of the measured/predicted corrosion depth read 1.01 and 0.92 for the mean end extreme corrosion depth values respectively, while the corresponding standard deviations read 0.38 and 0.23. If only ballast tanks are considered, then, the average ratios measured/predicted corrosion depth read 1.01 and 1.00 for the mean and extreme corrosion values respectively, while the corresponding standard deviations read 0.22 and 0.26. Therefore, it is concluded that the agreement between the predicted and measured corrosion depths is better for the ballast tank structural components than for the cargo tanks. One of the principal objectives of this thesis is to fit the existing nonlinear corrosion depth models to the corrosion measurements at 10th and 15th year and then to compare it with x iii the measurements at 20th year, aiming to investigate the uncertainty of the longterm corrosion depth prediction. The following conclusions are drawn: 1. The estimation of the ship hull girder section modulus loss after 20th year is satisfactory for 2 out of the 3 studied ships. Coating life is between 5 and 8.5 years. 2. The corrosion depth in cargo tanks is larger than the one in ballast tanks, while deck longitudinals in cargo tanks experienced the largest corrosion wastage of all structural elements analysed. 3. The nonlinear longterm local corrosion depth model based on the measurements after the 10th and 15th year in average predicts very well the corrosion depth after the 20th year. However, the uncertainty of that prediction, measured by standard deviation, is rather large. 4. There is no correlation between the corrosion severities of a local deck plates (longitudinals) and that of HGSM loss based on the corrosion models analyses. Deck plate can be severely wasted while the overall wastage of the entire hull section may remain at a very low level. Therefore, corrosion wastage of local plate and HGSM loss are to be treated independently. 5. Also, it was found that the new developed model is more accurate and reliable in ballast than in cargo tanks. 3. COLLAPSE STRENGTH OF PLATES AND STIFFENED PANEL SUBJECTED TO CORROSION AND FATIGUE CRACKS A model for collapse assessment of uniaxially compressed plates and stiffened panels has been developed and tested employing the nonlinear finite element method (NLFEM) using commercial software FEMAP with NX Nastran. The objective was to define a model that is simple enough to be applicable in everyday engineering practice. The principal difficulties are related to the choice of an appropriate definition of the model extent and boundary conditions, especially the mathematical description of the initial imperfections that have dominant influence on final collapse strength. The model for unstiffened plates is extended between the longitudinals along longitudinal plate edges and between the transverse web frames along transverse plate edges as has been shown in Figure 3.9. The typical element (mesh) size employed is 50 × 50 mm. This corresponds to about 16 elements in the shorter (transverse) direction and to about 80 elements in the longer (longitudinal) direction for an unstiffened plate of large oil tankers. x iv The boundary condition can affect significantly the plate strength. One bay plate model is used in this thesis, where the plate is simply supported at longitudinal edges and clamped at the transverse ones. This one bay model may be appropriate for uniaxially loaded rectangular plates. The extent of the model used for stiffened panel is one half of stiffener spacing on each side of the longitudinal in transverse direction and half of web frame spacing on each side of web frame in the longitudinal direction, as shown in Figure 3.10. For the present NLFEA the following boundary conditions are applied where T [x, y, z] indicates translational constraints, and R [x, y, z] indicates rotational constraints; ""0"" indicates constraints, and ""1"" indicates no constraints as is presented in Figures 3.13 and 3.14; Symmetric condition at AA0: R [1,0,0] with all plate nodes and stiffener nodes having an equal x  displacement; Symmetric conditions at BB0: T [0, 1, 1] and R [1, 0, 0]; Symmetric conditions at BB0: T [0, 1, 1] and R [1, 0, 0]; Symmetric conditions at AB and A0B0: T [1, 0, 1] and R [0, 1, 1]; At transverse floors: T [1, 1, 0] for plate nodes, and T [1, 1, 0] for stiffener web nodes. Welded metal structures during fabrications always have initial imperfections in the form of initial deformations and residual stresses [38]. The residual stress effect was excluded from the analysis in this thesis. Initial imperfections in elastic buckling mode with suitable magnitudes were chosen in the NLFEM analyses. The plate initial deformation magnitude is calculated by equations 3.3 and 3.4 and the result is presented in Figure 3.15. The shape of the initial deformation is obtained by linear elastic buckling analysis. For stiffened panels, three types of initial deformation are considered [37]: plate initial deflection, beamcolumn initial deflection, and sideways initial deflection of stiffener web. Shapes of the initial deformations are assumed to be similar to the stiffened plate buckling modes (equations 3.5 – 3.7). Total initial imperfections of the models are shown in Figures 3.16, 3.17 and 3.18. The model was tested by comparing NLFEM with two widely recognized methods for calculating the collapse (ultimate compressive strength) of unstiffened plates and stiffened panels. The first one is a simplified formula of harmonized rules for the construction of oil tankers with double hulls (CSR), and the second one is the PULS program of the Classification Society DNV. First, nine unstiffened plates made from three different types of materials were analysed and for each material three different thicknesses were chosen, as shown in Table 3.1 [38]. The plate slenderness, CE, was between 1.19 and 3.075, which represents a range of slenderness that usually occurs in practice. The panel dimensions were x v 4300 × 815 mm. Table 3.1 shows the results of ultimate strength obtained by NLFEM, program PULS, and Frankland's formula (CSR). The comparison between the results obtained by these three methods shows that a good match was achieved. Mean value of discrepancy of NLFEM compared to CSR is 3.9%, while the standard deviation reads 3.6%. The maximal discrepancy reads about 10%. This confirms the reliability of the used finite element model (NLFEM), as shown in Figure 3.23. Then, eight stiffened panels were analysed by NLFEM using the program Femap with NX Nastran. Three types of stiffeners were considered: flat profiles, Lprofiles, and Tprofiles. Geometrical characteristics of the stiffened panels, in millimetres, and the results of NLFEM to assess the ultimate strength are shown in Table 3.2. The ultimate strength σU for eight stiffened panels were in the range from 0.60 σY to 0.83 σY. The comparison of the results obtained using NLFEM, program PULS and the formula from CSR, for the abovementioned eight stiffened panels is presented in Table 3.3. There is a very good agreement of results obtained by the different methods. Mean value of the model uncertainty of NLFEM compared to CSR is 0.33%, while the standard deviation reads 4.52%. The largest difference between the results obtained using NLFEM and CSR reads about 7%, which is satisfactory. It is interesting to notice that the maximal difference between the two rule methods, PULS and CSR, is larger than the maximal discrepancy between proposed NLFEM and CSR. That is another favourable indicator of appropriateness of the NLFEM model proposed in the thesis. To conclude, it was shown that the proposed NLFEM model is effective, applicable, and that it can be used in practice for an evaluation of noncorroded, and uniformly corroded plates and stiffened panels. Advantages of NLFEM model in comparison with the analytical formulas of CSR rules and program PULSE is its ability to change a number of parameters and load conditions that occur during ship's service life. The cracks of plates, cracks of stiffened panels, corrosion in weld joints (grooving), and pitting corrosion often appear in the structure after several years of ship in service. These conditions can be simulated using NLFEM, and then analysed, which for the other two methods is not the case. Figures 3.26 to 3.47 present some of these phenomena and their effect on the strength of the panel deck structure. Differences in ""load–end shortening curves” as a result of nonuniform corrosion and cracks are presented in figs. 3.31, 3.36, 3.41 and 3.47. Definition of the simple structural model that is applicable in everyday engineering practice, its boundary conditions and initial deformations as well as verification of such model represents a contribution of the present thesis. Furthermore, there are no references in x vi the available literature describing the application of the FE software Femap with NX Nastran in collapse strength assessment of plates and stiffened panels. 4. HULL GIRDER ULTIMATE STRENGTH OF CORRODED SHIP STRUCTURE In this thesis, two methods for the assessment of the ultimate longitudinal capacity of hull structure are used. Both methods are in accordance with the requirements of CSR: the onestep method where PULS is used and the progressive collapse method using program MARS. The HGUS of three oil tankers is studied. The analysis is performed firstly for the initial asbuilt HGSM and then for reduced HGSM subjected to uniform corrosion, measured at 10th, 15th and 20th year of service. It was shown (Tables 4.7, 4.8 and 4.9 as well as in Figures 4.7, 4.8 and 4.9) that considering uniform corrosion wastage, the onestep method gives a 2% lower ultimate strength in comparison to the method of progressive collapse. The reason for this is that the method of progressive collapse gradually (in steps) calculates the limit of HGUS. It is assumed that once HGUS of individual deck stiffened panel is reached, loaded panel can still carry some load, before it comes to the point of the hull structure collapse. While the onestep method calculates HGUS from one step assuming that the collapse of the hull occurs as soon as bending moment reaches the limit value of the deck panels' strength. Also, based on the comparison of the effect of general (uniform) corrosion for all three models (ship 1, ship 2, ship 3), it can be concluded that after the hull surveys at 10, 15 and 20 years there was not any reduction in the longitudinal strength of more than 10%, that is acceptable reduction in accordance with the IMO requirements. In all three ships this 10% reduction could happen between 25 to 30 years, i.e. later than designed ship lifetime. The achieved results of collapse strength of plates and stiffened panels, described in chapter 3 of the thesis, are incorporated in the calculation of HGUS. The results presented in Figures 4.10 to 4.13 confirmed that the degradation effects, such as uneven corrosion wastages (grooving and pitting) and cracks, could significantly reduce the ultimate strength of hull structure. Research is performed within the present work in order to find appropriate way to incorporate aging effects in HGUS calculation. For the case of nonuniform corrosion and fatigue cracks it is proposed to combine the methods for HGUS with NLFEM of single deck panel. x vii 5. STRATEGIES FOR SURVEY, MAINTENANCE AND REPAIR IMPROVEMENT OF AGING SHIPS IN SERVICE In the last chapter of this thesis several practical approaches are proposed to assist the process of survey and maintenance of ships in service. First, the method for predicting the corrosion degradation can be presented in the form of corrosion maps, allowing surveyors to facilitate the practice of making costly repairs and detailed examination of the hull structure, as presented in Figures 5.1 to 5.4. The allowable thickness reduction for deck structures of oil tankers reads 4 mm, with the reference to CSR. This value is the margin of the comparison considering actual corrosion wastages. In a current research project within the IACS harmonization project team on CSR corrosion addition, a value of 4 mm is also considered as acceptable [12]. Then the increase of deck stress due to the loss of HGSM was analysed and at the same time the reduction of the collapse strength of deck structure due to the negative effects of corrosion was determined. This simple method (see Figures from 5.5 to 5.16) shows the point in time when it is expected that the applied stress will exceed the strength of stiffened panels of the main deck. Classifications societies then suggest replacement of the corroded plating with a new one in order to avoid an unsafe zone for the aging ships. This approach is for the first time proposed by Gua et al. [10] for unstiffened deck plates, while it is extended in the present thesis to the stiffened panels including the effect of uneven corrosion (grooving and pitting) and cracks (presented in Figures 5.17 to 5.20). These calculations are based on methods and results developed in the 3. chapter of this thesis. The vertical bending moment, as one of the most important criteria for longitudinal strength, must be controlled during the ship's service. However, such calculations are subjected to uncertainty regarding the unpredictability of the corrosion processes and wave loads. To account for uncertainties, the classification societies use partial safety factors. These factors are determined using probabilistic methods and practical experience from the past and their values can vary significantly. To demonstrate these large differences, in this thesis two ""sets"" of partial safety factors were used according to the rules of BV and CSR. Comparing the results, it was observed that in all three ships and their respective hull conditions (after 10, 15 and 20 years) HGUS is in accordance with the requirements of BV rules for the uniform corrosion, while the CSR results are considerably worse (presented in Figures 5.20 to 5.30). In calculation of HGUS, results obtained in the 4. chapter of the thesis are employed. Finally, an improved method for estimating the fatigue strength of deck structure of oil tanker is shown. The employed fatigue calculation is similar to the CSR method. The main x viii difference from the CSR approach, where the constant wastage is used from the beginning to the end of the ship's life, is that the realistic corrosion degradation mode, developed in the 2. chapter, was used in the present study. The accumulated damage is calculated for each year of service individually, taking into account the actual stress of HG, determined based on the measured HGSM loss. The results of these calculations, for three ships, are shown in Figure 5.31, and Tables 5.15.3. The tables also show the results for a constant HGSM loss over the entire lifetime according to the fatigue requirements from CSR rules (25% of the maximum corrosion of particular hull elements) and also for longitudinal strength requirements (50% of the maximum corrosion of particular elements). The a.m. tables show that the calculated lifetime according to the measured corrosion wastage may give better insight into the hull girder fatigue behavior compared to the rule approach. 6. CONCLUSION AND GUIDELINES FOR FURTHER RESEARCH The researches which are described in this thesis are aimed to improve the application of direct calculations in the planning of survey and repairs of ship's structure. Further research in this area can be recommended within two fields: đ Development and practical implementation of crack propagation methods to the ship's structure; đ Development and application of reliability and riskbased methods for maintenance of ship structures. Crack propagation methods based on fracture mechanics are currently being developed Benefit of using these methods is that small fatigue cracks that appear around the welds can be tolerated and analysed. The crack propagation methods have still not found appropriate place in the shipbuilding industry because of the inherent complexity of welded ship structures. Consequently, there are a number of uncertainties related to the determination of their load and capacity and also regarding the variability of the effects of corrosion. Especially important are the effects of various parameters that can affect determination of the stress intensity factor. A typical example is the uncertainty in calculation of the stress concentration factors in presence of the cracks. Regarding fatigue load of ship structures, the greatest difficulty is the unpredictability of random wave loading, which is a major factor in fatigue loading and crack propagation. The prediction of crack growth under conditions of random loading is a challenging and complex problem that is currently the subject of research. Improvement of the methods for the analysis of the crack propagation could have an extremely p"
Item Type:  Thesis (Doctoral thesis) 

Uncontrolled Keywords:  tanker za prijevoz nafte; korozijska istrošenja; zamorne pukotine; inspekcijski pregledi; ukrepljeni panel; kolapsna čvrstoća; nelinearna analiza metodom konačnih elemenata; granična čvrstoća brodskog trupa 
Keywords (Croatian):  oil tanker; corrosion wastage; fatigue cracks; hull structure survey; stiffened panel; collapse compressive strength; nonlinear finite element method; hull girder ultimate strength 
Date Deposited:  22 Sep 2014 18:00 
Last Modified:  16 Oct 2015 12:57 
URI:  http://repozitorij.fsb.hr/id/eprint/2081 
Actions (login required)
View Item 