VALIDATION BY CLASSIFICATION SOCIETES OF LETS GLOBAL LIFE EXTENSION PROCEDURE FOR OFFSHORE INSTALLATIONS

Transcription

1 VALIDATION BY CLASSIFICATION SOCIETES OF LETS GLOBAL LIFE EXTENSION PROCEDURE FOR OFFSHORE INSTALLATIONS Luis Lopez Martinez LETS Global Postbus JA Rotterdam, The Netherlands Abstract LETS Global has during five years developed its own procedure and equipment for fatigue life extension of offshore structures by ultrasonic peening treatment. The general validation of the procedure requires specific knowledge about the consequences on the UP improvement when compressive stresses are relaxed during the remaining service life of the installation. Fatigue testing has been carried out in Norway and Sweden where fatigue loading sequences have been specifically designed to comply with the requirements of Classification Societies. The fatigue test specimens have been subjected to a sequence of static loads previous to fatigue testing to ensure almost complete relaxation of compressive residual stresses induced during ultrasonic peening treatment. The results demonstrated that the degree of life extension could be retained even if the treated structure would suffer the most severe loads conditions during its remaining service life after ultrasonic peening treatment was applied. Subsequent fatigue test series consisted of welded and UP-treated cruciform specimens where compressive stresses have been drastically decreased by cutting them to a reduced width. These fatigue test results confirms the previous UP test results where compressive residual stresses were relaxed and the improvement was retained. The fatigue life extension achieved for improved welds is four times calculated on the basis of FAT curves when compared to its endurance in as-welded condition for a given welded detail. The degree of improvement is up to ten times when high cycle regime is considered. 1 INTRODUCTION The application of fatigue life improvement techniques is gaining popularity in the last years. Classification Societies have been focusing more and more on these and the latest document dealing with it [1] presents recommendations for weld toe profiling by machining and grinding, weld toe grinding, TIG-dressing and hammer peening. The other important document in respect to execution of the improvement is the IIW Recommendations [2], which contains extensive reference data for various fatigue life improvements [10] and the quality assurance and control of their application. Fatigue life improvement techniques can contribute to reduce maintenance cost by the avoidance of recurring weld repairs. Furthermore life extension techniques are the only remedy when higher stress and/or fatigue cracks occur in a structure with many years remaining service life. One of the most promising techniques currently used to extend the fatigue life of welded joints is ultrasonic peening. The technique consist of introducing a weld toe grove at the weld toe, cleaning the weld toe from possible crack prone sites as well as introducing compressive stresses during the same and solely working operation. The effect of cleaning weld toe region as well as introducing a weld toe groove have undisputed benefits which are also more easy to quantify. However it still exits two main concerns in reference to the long term quantification of the fatigue life extension achieved by ultrasonic peening : a) It is a well known fact that the type of spectrum or the distribution of loading history in compression and in tension has an effect on Miner s rule [3] and this effect may also apply for improved welds. How the loading history interacts with the compressive stresses produced during UP treatment, which is the case in all treated real structures, underline the necessity of spectrum fatigue test or specially designed constant amplitude fatigue testing of UP treated welds.. b) If the compressive stresses produced by UP treatment would be relaxed by external loads suffered by the structure during its remaining service life could we still count on the degree of improvement assessed on the basis of constant amplitude fatigue test results from laboratory? 1 of 11

2 Fig 1 Ultrasonic Peening Treatment Some recommendations do exists for a) in literature [3]. However, it is not very much done for b). Further more and due to the profound implications of any miscalculation of b) it is of paramount importance to have a better understanding of the subject before any general recommendations can be given by Classification Societies. Therefore LETS Global as a leading expert company in the field have been actively working to develop fatigue testing procedures which will address the issue and brings some light to this vast area of the fatigue knowledge. 2 PREVIOUS INVESTIGATIONS ON THE INFLUENCE OF SPECTRUM LOADING ON RESIDUAL STRESSES 2.1 Spectrum loading sequences investigated to assess the general effect on residual stress relaxation The influence of various spectrum parameters on residual stress relaxation have been studied for specimens in the aswelded condition [5]. This information contributes to the basic understanding of the residual stress relaxation process as it happens during variable amplitude fatigue loading, which is the type of loading for the majority of fatigue loaded structures. Fig 2 Range-pair distribution for six different spectra. The different spectra represented a wide range of parameter combinations in terms of Mean Stress, Stress Ratios, Irregularity Factor (number of positive mean crossings divided by the total number of cycles in the block) and p which is the ratio of minimum load to maximum load in an exceedance distribution. The Range pair distribution for the six different spectra used is presented in Fig 2. 2 of 11

3 2.2 Relaxation of residual stresses during spectrum loading in as-welded specimens One further step necessary for the understanding of the residual stresses relaxation process is the study of the relaxation of weld induced residual stresses (normally tensile stresses) due to spectrum fatigue loading. The effects of different spectrum loading on weld induced tensile stresses relaxation is presented in Fig 3. The percentage of relaxation is plotted against percent of total fatigue life. Total fatigue life means for the current fatigue test series number of cycles until failure. Fig 3 Residual stress relaxation versus percentage of total fatigue life Fig 3 shows that within 8% of the total fatigue life of the specimen the relaxation of tensile residual stresses is 50% or more of its initial value. These results concerns only specimens in the as-welded condition but it should be reasonable to accept that for various spectrum parameter relaxation (or redistribution of residual stresses) occurs at a very early stage of the fatigue life of a component. It remains to be clarify if the rate of relaxation for compressive stresses is similar to the rate of relaxation of weld induced tensile stresses as reported in Fig 3. One argument which would influence the rate of relaxation is the geometrical stress concentration at weld toe, which in UP treated weld is considerable less than in the as-welded condition. The consequence of a lower stress concentration in treated welds could be a higher percentage of the total fatigue life to achieve same level of redistribution of residual stresses. 2.3 Spectrum loading relaxation of compressive stresses induced by fatigue life improvement techniques The influence of the different spectrum parameters on residual stresses for TIG-dressed specimens has been also extensively studied in [8]. Despite the fact that the level of induced compressive stress produced by ultrasonic peening differs from those produced by TIG-dressing at the weld toe zone, we can have a indication of relaxation pattern if we study in detail how spectrum loading redistribute the stresses during fatigue loading, see Fig 4. As explained in 2.2 geometrical stress concentration at weld toe region might have an influence on the rate of residual stress relaxation also under spectrum loading. As a first approximation and based on [10] and as fatigue test results for high stress ranges for TIG- and UP-improved welds are similar we could assume similar, although no exactly the same, geometrical stress concentration after treatment for those. The TIG-dressing induced compressive stresses are in the order of 200 MPa, (difference red and blue lines at weld toe in Fig 4), from 500 MPa in the as-welded condition to 300 MPa after TIG-dressing. The application of spectrum load contributes to a clear and progressive redistribution of residual stresses. It is apparent from Fig 4 that the loading history in compression and tension interacts with the residual stress field producing at least for the studied spectrum a more even general distribution of internal stresses. Furthermore it is important to keep in mind the level of original stresses are influenced by parent plate and weld metal yield strength as well as welding procedure, the diagram in Fig 4 shows only a general shape of redistribution of residual stresses during spectrum loading. 3 of 11

4 Fig 4 Spectrum relaxation of TIG-dressing induced residual stresses (Steel grade: Domex 590) 3 RESULTS OF CONSTANT AMPLITUDE FATIGUE TESTING OF PRELOADED SPECIMENS On e way to simulate the effects of spectrum loading sequences on the redistribution of compressive stresses is to apply static preloading to fatigue test specimens previous to the constant fatigue testing. In order to investigate the effect of compressive stresses on the improvement from ultrasonic peening it has been suggested [4] to apply some compressive stresses ranges into the test specimen at the start of each fatigue test. This is based on the fact that the largest compressive stresses a real structure would experiment during its service life, is approximately the nominal stress range. The following fatigue test series have been proposed by Classification Societies in order to address to consequences of a relaxation of compressive stresses (UP induced) on the fatigue life extension: - Nominal Stress in Compression equal material yield strength: see fatigue test results below - Nominal Stress in compression equal 0.5 material yield strength; - Without pre-stress; - Nominal stress in tension equal 0.5 material yield strength. Furthermore it is recommended as well to carry out the constant amplitude fatigue test with large R value. 3.1 Fatigue test specimens The first fatigue test series carried out to investigate the effect of compressive stresses relaxation on the degree of improvement have been reported previously [7]. The preloading sequence for this fatigue test series was designed according to DNV advice and it is part of complementary tests dedicated to highlight the effects of compressive stresses relaxation. Fig 5 Fatigue test specimen 3.2 Preloading sequence Preloading sequence consisted in five times loading in compression to 0.85 yield strength of material. After these five cycles the specimens where subjected to constant amplitude fatigue loading R=0.1. The preloading sequence is shown in Fig 6. 4 of 11

5 Fig 6 Preloading sequence for three point bending specimens Of the cases proposed in 3 this preloading sequence could be considered severe. In other words it is likely that most of the compressive stresses induced by UP treatment have been relaxed or at least completely redistributed after the preload sequence have been applied to every specimen. Such a severe external load in a structure in service it is not very common. 3.3 Fatigue test results The evaluation of fatigue test result is presented on the basis of SN-Design Curves. All comparisons have been done against mean-curve minus two standard deviations. The fatigue test results for UP treated Class F specimens are compared to hammer peened specimens. A FAT Class has been calculated for every fatigue test series presented in Fig 7. The summary of the results is presented in Table 1. The values for TIG dressed and ground are from reference [1]. AW Ground TIG HP UP FAT [MPa] m Life factor n/a Strength factor n/a Table 1 Summary of fatigue test results; HP and UP preloaded in compression to 85% of YS 5 times. Fig 7 Fatigue test results for UP-treated weld after 5 preloading cycles at 85% of yield stress 5 of 11

6 The compressive preloading applied to the ultrasonic peening treated specimens do not noticeable reduces the expected life extension of the treated welds. The influence of the preloading has been extensively studied [5] and [6] in connection to spectrum loading of tension/compression type for TIG-dressed specimens where same effect has been detected. 4 RESULTS OF CONSTANT AMPLITUDE FATIGUE TESTING OF SPECIMENS WITH REDUCED WIDTH Other way to achieve almost complete relaxation of compressive stresses induced by UP treatment is to machine the fatigue test specimens to a reduced width. The fatigue test specimens in Fig 8, width=10 mm, have been tested under constant amplitude with R=0.1. The fatigue test results for UP treated specimens differs from the fatigue test results reported in Fig 9 by approx 30% and that is due to the be difference between bending fatigue test- and tension fatigue test results presented in Fig 9. Fig 8 Fatigue test specimen width 10 mm 4.1 Fatigue test results Fig 9 shows fatigue test results for specimen in Fig 8 [14]. The slope for UP treated specimens, natural mean curve, shows a less steep slope if compared to normal slope (m=3) for as-welded specimens. It would be an indication that residual stresses induced by the UP treatment have been almost completely relaxed due to machining of specimens to reduced width. Furthermore, it would be noted a significant improvement for the UP treated specimens when compared to characteristic curve for as-welded specimens (red curve) for this specific weld detail σ r [MPa] 100 failures run outs natural mean curve mean curve( m=3) char. curve 10 1,E+04 1,E+05 Cycles 1,E+06 1,E+07 Fig 9 Fatigue test results for specimens 10 mm width 5 ULTRASONIC PEENING TREATMENT OF SPECIMENS WHICH HAS ALREADY CONSUMED HALF OF ITS FATIGUE LIFE The life extension of steel structures currently in service is the main area of application of the LETS Global procedure. As a result all the structures we are currently treating with ultrasonic peening have consumed at least half of its design fatigue life in service. Therefore it is important that to document how the ultrasonic peening affects or 6 of 11

7 improve the fatigue resistance of welds which has consumed half of its design fatigue life. Fatigue test of specimens both under constant amplitude and under spectrum loading will be treated by ultrasonic peening after half of their estimated fatigue life is consumed. However, in practice it is necessary to take into account that some of the weld connections proposed for treatment could have fatigue cracks already. The options we face in practice then could be as follows. 5.1 Weld connections in service which does not shown during NDT fatigue cracks The LETS Global ultrasonic peening procedure for life extension starts with a NDT examination. This is to ensure that very deep fatigue cracks will not be embedded in the weld. This NDT examination allows as well for judgement of basic weld quality and if a treatment of weld reinforcement would be necessary. If no cracks are found, or only small cracks are found and weld quality is good, the ultrasonic peening can be carried out straight away according to our own developed procedure. However, if deep cracks are found one of the following alternatives can be followed. 5.2 Weld connections in service which shows fatigue cracks and or bad weld quality A weld connection under high stress in service during a long period of time could show fatigue cracks when we apply NDT and previous to UP treatment. The relation of the exposed crack depth in relation to plate thickness is an important parameter. If plate thickness is 20 mm and a crack depth 2 mm UP treatment could restore the fatigue strength [11] up to its original design level. This assumption is based on previous fatigue test results of hammer peened welds which contained cracks before the application of hammer peening. Based on our fatigue test results Fig 7 and its comparison to hammer peened welds it would be conservative to assume that the same crack depth could be treated with UP and achieve similar results as reported in [11]. 5.3 Weld connections in need of weld repair before application of ultrasonic peening Fig 10 Ultrasonic peening treated weld connection after replacement and weld repair Some cracked weld connections are in an unavoidable need of repair. Furthermore it is advisable in some cases to improve the geometry, see Fig 10, of structural components so at the transition of stresses in the structure is improved. The repaired weld in these re-shaped structural components is subsequently UP treated with the normal UP procedure, see 8. 6 STATE OF THE PROJECT 6.1 Fatigue testing The following fatigue test matrix shows the test program designed to clarify the fatigue behaviour of UP treated welds subjected to variable amplitude fatigue loads, specifically to allow for the relaxation/redistribution of compressive stresses. 1. Constant amplitude 2. Constant Amplitude with preloading 3. Spectrum with compressive overloads UP after half as-welded endurance achieved UP after half as-welded endurance achieved UP after half as-welded endurance achieved 7. o denotes a fatigue test series done and x denotes a test series which are being/will be carried out 7 of 11

8 AW o o x x x x AW+UP o o x x x x Table 2 Fatigue test series designed to cover compression relaxation effect on UP treated welds The aim is to demonstrate that the relaxation of compressive residual stresses do not noticeable diminish the degree of fatigue life extension assessed for a specific weld connection treated by UP. Furthermore if any specifically load sequence, meaning combination tension/compression, would almost completely relax the compressive stresses what would then the consequence be for the long term fatigue life improvement? 7 LETS GLOBAL LIFE EXTENSION PROCEDURE APPLICATION Fatigue life extension by ultrasonic peening has been achieved in high stressed areas; see Fig 11, on structural details of an offshore installation. The treated welds were located in several frames in the cargo deck. Some of the treated frames have been weld repaired previous to the application of the ultrasonic peening.. The high stressed areas requiring special attention were the deck stool weld joints at bracket side. These showed to be more critical than the bracket free edge. These critical areas, where high stresses and varying weld quality come together, were selected for the post-weld treatment. Fig11 High stress areas and ultrasonic peening treatment (FEA courtesy of Bureau Veritas) The fatigue life assessment of these structural details was based on principal stress plots. This assessment showed too short fatigue lives for the relevant welds before treatment and the aim with the ultrasonic peening treatment was to avoid any further weld repairs during the remaining service life of the installation at these specific locations. 7.1 Ultrasonic Peening Treatment Procedure The procedure for the treatment consists normally of three steps described below. Depending on the original weld quality, weld spatter, access to previous NDT inspections documents, etc the time needed for treatment can differ. a. Preparation of weld toes: done with a 3 mm pin diameter. Ultrasonic treatment promotes the loosening of any impurity located at weld toes including inter-pass weld toes. See Fig. 12 Fig 12 Weld toe and reinforcement; free of weld defects, impurities and cold-laps. 8 of 11

9 b. Weld groove creation: done with a 4 mm pin diameter, see Fig 13. The final radius at the weld toe 2 mm and groove 0.5mm and show clean metal surface after the treatment. Fig 13 Creation of weld toe groove: reduction of stress concentration c. Treatment of weld reinforcement: normally done with a multi-striker working head. See Fig 14 below. Fig 14 Treatment of weld reinforcement or weld flange with multi-striker. 7.2 Quality Control & Quality Assurance The achieved geometrical weld parameters for the treated welds are presented in Table 3. The parameters are part of the Ultrasonic Peening Procedure and ensure that the improvement achieved in previous treated details, is exactly reproduced in the structural members treated in the installation. Table 3 presents the mean measurement values for all welds treated. Weld flange angle mean [º] Weld toe radius mean [mm] Weld grove deep mean [mm] Weld flange surface quality Before treatment 45 o n/a n/a Weld spatter, start/stop, etc After treatment 45 o Smooth, shiny 8 LITERATURE RESULTS 8.1 Fatigue test results Table 3 Geometrical weld parameters for welds before and after treatment In reference [12] the results from Fig 7 have been compared to similar fatigue test. The results from the preloaded fatigue test results, Fig 7, appears to be lower that the specimens which have not been subjected to static preloading [12]. That indicates it could be an effect which needs to be taken into account when assessing the long term level of improvement achieved by UP treated welds. However, it is a major uncertainty in the correlation done in [12] since results of preloaded specimens, Fig 7, refers to 350 MPa steel grade whereas the results in [12] refers to 700 MPa steel 9 of 11

10 grade. As well understood [10] the parent yield strength plays a roll in fatigue strength when welds are treated by improvements techniques and specifically UP. 8.2 Residual stresses. The level of compressive stresses induced by various ultrasonic treatment techniques has been found to be in the order of -134 MPa at weld toe region for 355 steel grades [13]. As the same grade of steel has been used for the static preloaded fatigue test series reported hereby it is almost certain that the UP induced compressive residual stresses were completely redistributed and/or relaxed by the applied preloading sequence reported in [7]. 9 CONCLUSIONS For long term assessment of fatigue life improvement by ultrasonic peening it is necessary to understand the influence of compressive stresses relaxation on the degree of improvement. Spectrum loading redistributes the residual stresses at a relative early stage of the total fatigue life. This fact could be valid even for welds treated by ultrasonic peening. Studies of relaxation of TIG dressed welds under spectrum loading showed that a redistribution of by the treatment induced stresses occurs. This redistribution contributed to keep the residual stresses at a low value at weld toe region for high cycle fatigue. Assuming similar geometrical stress concentration for treated welds, TIG and UP, similar effect would occur for UP treated welds as a conservative estimate. UP treated welds which were subjected to severe static preloads previous to fatigue testing achieved a FAT value of 151 MPa which is 100% improvement when compared to as-welded condition. These results confirms the previous conclusions based on literature results. Cruciform fatigue test specimens manufactured to reduced width and therefore with almost no compressive stresses showed retained fatigue life improvement. QA & QC of ultrasonic peened welds is easy to control in terms of weld toe radius, weld toe grove and weld reinforcement surface quality. 10 REFERENCES 1- DNV-RP-C203 (2005) Fatigue Design of Offshore Steel Structures, August DNV 2- Haagensen P.J. and Maddox S.J. (2004) IIW Recommendations on Post Weld Improvement of Steel and Aluminium Structures ; Haagensen P.J. and Maddox S.J. (2009) IIW Doc. XIII The Fatigue Behaviour of Steel Structures under Random Loading; Agerskov H. IIW Doc. XIII Direct communication with Inge Lotsberg, DNV 5- Spectrum Fatigue Testing and Residual Stress Measurements on Non-Load carrying Fillet Welded Test Specimens; Bogre J and Lopez Martinez L.; Proceedings of Fatigue Under Spectrum Loading and in Corrosive Environments, August 1993, The Technical University of Denmark, Lyngby, 6- Influence of Spectrum Loading on the Fatigue Strength of Improved Weldments; Lopez Martinez L. and Blom A.F. (1997) International Conference on Performance of Dynamically Loaded Welded Structures. Editors S. J. Maddox and M. Prager; IIW 50 th Annual Assembly Conference 7- Life Extension of Class F and Class F2 Details Using Ultrasonic Peening; Lopez Martinez L and Per J. Haagensen (2007) IIW Doc. XIII Investigation of Residual Stresses in As-Welded and TIG-Dressed Specimens Subjected to Static Load and Spectrum Loading ; Lopez Martinez L. et al. ; Proceedings of the First North European Engineering and Science Conference (NESCO I) Subtitled "Welded High Strength Steel Structures" held in Stockholm, Sweden, 8-9 October Comparison of Post Weld Treatment of High Strenght Steel Welded Joints in Medium Cycle Fatigue; Pedersen M. M et al. IIW Doc. XIII Fatigue Behaviour of Welded High-Strength Steels, Paper D; Lopez Martinez L. Report 97-30, Department of Aeronautics; The Royal Institute of Technology; SE , Stockholm, Sweden Life Improvement and repair of Fatigue Cracks in Welded Joints by Hammer Peening; Per J. Haagensen, Paper M, OMAE Comparison of Post Weld Treatment of High Strength Steel Welded Joints in Medium Cycle Fatigue; Pedersen M.M. et al. IIW Doc Xiii of 11

11 13- Determination of Residual Stress Depth Fields in Welded Joints After Different Mechanical Surface Treatments; Nitsche-Pagel Th. et al. Institute of Joining and Welding, Technical University of Braunschweig, Langer Kamp 8, Braunschweig, Germany 14- Fatigue Testing of Cruciform Joints for Different Weld Positions; Z. Barsoum, KTH Lightweight Structures, Teknikringen 8, SE Stockholm, Sweden 11 of 11