Buckling Failure of Slender Tubular Member of Offshore Structure with Cutout Presence

Transcription

1 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 11 Buckling Failure of Slender Tubular Member of Offshore Structure with Cutout Presence Miftahul Iman, Bambang Suhendro, Henricus Priyosulistyo, and Muslikh Abstract Slender steel tubular members are widely used in jacket platform truss structures. When the member is in compression, buckling phenomenon governs its capacity. The critical load (Pcr) is a main indicator of buckling failure. Euler formula for predicting the capacity of slender compression member is no longer applicable when defect, such as a hole resulting from pitting corrosion, presence on the member. This research was conducted to study experimentally the effect of a circular hole on the buckling capacity of slender steel tubular member commonly found in typical platform structures. Variation on hole positions was at 0.15 L, 0.5 L, and 0.5 L, where L is the length of the member. The hole was 0.5 pipe diameter. Experiments were carried out in two scales, namely small models represented by seamless pipe SCH40 with ½ inch diameter, and large models that used similar pipe with diameter of inches. Both models had the same slenderness ratio of 16. Monotonic axial compressive loading was applied to the member up to buckling occur. Load cell was used to monitor the applied load, while the axial and lateral displacements at the center were observed by LVDT s. Strain gauges were placed closed to the cutout position. From the test results upon 1 specimens it was observed that: (a) the presence of cutout reduced the buckling load significantly, (b) the reductions ranging from 3 to 1%, depending on the hole posistions, (c) the maximum reduction occurs when the hole position was in the middle of the member length, and (d) the buckling modes corresponded with member bent away to opposite side of the cutout position. Pitting corrosion is a form of extremely localized corrosion that leads to the creation of small holes in the metal, usually initiated by defect or local delamination of the protective coating. It is common that compression members on a jacket structure has a slenderness ratio (l/r) of more than 10. Several of them are found in the prototype of platform structures located in East Kalimantan, Indonesia (Table I). The Euler formula, commonly used for predicting the capacity of slender compression member P cr EI (kl) where P cr, E, I, k and L is respectively the critical buckling load, modulus of elasticity, cross sectional moment of inertia, effective length factor which depends on the end support conditions, and length of the member, is no longer applicable when defect, such as a hole resulting from pitting corrosion, presence on the member [1]. Additionally, buckling formula found in many standard practical codes do not consider the presence of defect such as a hole in the compression member. The effect of holes is only considered in determining the capacity of tension members. Index Term pitting corrosion buckling, tubular member, cutout, offshore, Pn FcrAg () I. INTRODUCTION Slender tubular members are used widely in civil engineering structures, such as trusses for highway bridge, roof supporting system, and offshore platform. When the member is in compression, buckling phenomena governs its capacity. In offshore structures, as well as other structures located in marine environment, pitting corrosion is commonly occured and potentially reduces buckling capacity of the member, and in turn reduces the overall structural capacity. Miftahul Iman, S.T., M.Eng. Doctoral Candidate in Civil Engineering Sciences, Faculty of Engineering, Gadjah Mada University, Indonesia Phone: , anmifal@yahoo.com Bambang Suhendro, Ir., M.Sc., Ph.D. Professor Department of Civil and Environmental Engineering, Gadjah Mada University, Indonesia. bsuhendro@ugm.ac.id Henricus, Priyosulistyo, Ir., M.Sc., Ph.D. Professor Department of Civil and Environmental Engineering, Gadjah Mada University, Indonesia. priyo_ugm@yahoo.co.id H. Muslikh, Dr., Ir., M.Sc. M.Phil. Associate Professor Department of Civil and Environmental Engineering, Gadjah Mada University, Indonesia. muslikh7@yahoo.com Equation () had been used for determining the nominal load (P n) of a compressive member [], where F cr is critical stress which calculated by two diffrence states. They are elastic and inelastic buckling states. The conditions had been distinguished by the member effective slenderness ratio (kl/r). Research on buckling of tubular steel shells with circular cutout has been investigated [5]. The experimental buckling tests have been conducted using a Servo-hydraulic machine (Instron 880). The entire sample set as a short column (l/ r 50). The material used in the study was 316Ti stainless steel. Properties of the material used were E = 187 GPa, F y = 334 MPa and Poisson's ratio In this study, the influence of shell length, shell diameter, shell angle and diameter of circular cutouts on the predicted buckling values for the tubular shell has been explored. Numerical simulations of tubular shell subjected to combined loading were conducted. The analytical solutions show excellent agreement with the numerical results predicted by FEM.

2 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 1 Study on buckling of tubular member with various materials and shape of cutouts have been conducted. Numerical and experimental investigations of the response of aluminium cylinders with a rectangular cutout subject to axial compression had been reported [3], as well as numerical and experimental investigations on buckling of steel cylindrical shells with elliptical cutout subjected to axial compression [4]. The following empirical formulas, respectively, for predicting reduction coefficient of the buckling load with the presence of cutout were resulted from those studies. The buckling reduction factor of compression member with a square shape hole [3], K d 3 3 3,, A B C D E F G H I J K L M N a O P Q ormula is, however, suitable for small and intermediate slenderness ratio of compression member. For elliptical cutout, the followin is applied [4],, A B C D E F K cutouts where = L/D and = L 0/L. The effect of hole locations as well as hole sizes in the buckling capacity of slender steel tubular member has been preliminary investigated [1]. The study was conducted on cilcular cross section steel tubular members with outer diameter of inches and slenderness ratio of 16, by both experimental and numerical approaches. The preliminary results showed that the critical buckling loads tend to decrease when position of the holes, respectively, at the 0.15, 0.5, and 0.5 length of the member, compared relatively to the intact tubular member. Simulation and analysis of steel cylindrical shells with various lengths, including quasi elliptical cutout, subjected to axial compression load were systematically carried out using finite element method and the investigation examined the influence of the cutout location and the shell aspect ratio on the buckling, and the postbuckling responses of the cylindrical shells. For several specimens an experimental investigation was also carried out via an INSTRON 880 servo hydraulic machine [11]. The results obtained from the experiments were compared with numerical results. A very good agreement was observed between the aforementioned results. Finally, corresponding to experimental and numerical results, an expression was derived for finding the buckling load of such structures. The effect of opening in thin cylindrical shells under compression loading has been investigated considerably. However quite a few of the references are directly related to the cutouts along the length of the shells in the form of (3) entrance door. Different buckling modes as well as the effect of geometric parameters of a cutout along the length of the tubes with door-shaped cutouts under axial loading has been examined and reported [9]. A stiffening method was used to decrease the effect of the cutout on the capacity of such structures. The influence of imperfections on the buckling behavior of aluminum cylindrical shells had also been studied both experimentally and numerically [13]. Imperfection was provided in the form of one or two circular holes on the surface of the shell. Specimen configuration was determined by the geometric parameters α, which was calculated based on the radius of a circle hole, thickness t, and the cylinder radius R. In this experiment, Digital Image Correlation Method (DICM) was used to determine the strain and 3D displacement during testing. MSC Marc Mentat 005 computer software was used for numerical analysis. The results showed that the effect of imperfections was very sensitive to the buckling load of specimens. The results indicated a good correlation between the experimental results and the finite element method. This research was conducted to study experimentally the effect of a circular hole on the buckling capacity of main slender steel tubular member with slenderness ratio, L/r, of 16 which commonly found in typical platform structures in Indonesia. Variations hole layout was determined at 0.15 H, 0.5 H, H, and 0.5 H, where H is the length of the compression member. The hole was 0.5 pipe diameter, and the pipe was made of SCH40. II. SPECIMEN PREPARATION Specimen preparation consisted of material selection, material properties tests, and experimental modeling. A. Material selection and properties tests SCH seamless pipes which meet the standards of API 5L/A53/A106 Grade B has been selected as a material model in this study. The modulus of elasticity (E) as well as yield stress of the material were tested by means of tensile test (Fig. 1), while the Poisson's ratio was taken to be 0.33 []. The support system was chosen to be pinned type using JIS S45C medium carbon steel material, as shown in Figure. Fig. 1. Tensile test specimens The ASTM standard for tensile and flexural tests have been connducted to determine the modulus of elasticity (E), and

3 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 13 yield stress (f y) of the material [7]. The results were MPa for E, MPa for f y (small model material), and 40 MPa for f y (large model material). B. Experimental modeling The experimental model had been derived based on the Buckingham Pi Theorem in dimensional analysis. The prototype was a diagonal element of typical jacket platform structure located in East Kalimantan, Indonesia, having geometric as well as material properties listed in Table I [9]. SECTION A (mm) TABLE I PROTOTYPE DATA GEOMETRIC PROPERTIES OD ID t (mm) (mm) (mm) D Based on the Buckingham Pi Theorem, and realizing that the critical buckling load (P cr) of a slender tubular compression member with a circular cutout depends on the effective length of the member (l), cross sectional moment of inertia (I), elastic modulus of the material (E), outer diameter (D) and thickness (t) of the tubular member, diameter of the hole (d), and the distance of the hole (l 0) measured from the end of the member, the following equation could be formulated L/r D/t slenderness ratio of 16. The geometric scale for small model (S l) was taken to be 3, while that of the large model was 10. The material scale (S E) for both models were taken to be 1. The small model had internal diameter mm, thickness.6 mm, length 600 mm, and fy = MPa, while the large model had internal diameter 5.70 mm, thickness 3.90 mm, length 000 mm, and fy = 40 MPa. Both material models had the same modulus of elasticity E = MPa. III. EXPERIMENTAL TESTS A. Small model The experiment was conducted on four seamless tubular pipe SCH40 with the length of 600 mm, ½ inch diameter, and slenderness ratio of 16. Circular holes with diameter of 0.5 pipe diameter were provided respectively at 0.15L, 0.5L, and 0.5L, where L was the length of the specimen. One specimen without hole was also tested for reference. It was intended to be preliminary tests before conducting experiments with the relatively larger models. The pipe specimen was pinned supported at both ends (Fig. ) and monotonically loaded with Universal Testing Machine up to buckling occur. Dial gauges and strain gauges were used to monitor displacements and strain, respectively. Before being tested, all tubular pipes were straightened and perimeter controlled [8]. Buckling test setup for small model is shown in Fig. 3. P cr E d l 0 I C 4 l 0 j a D d l l 0 b t d c (5) where C is a constant derived from the experiments. The equation was then rewritten in terms of nondimensional variables п i (i = 1,, 3, 4, 5) as follows: (6) j a b c 1 C To satisfy all similarity requirements between prototype and model, п i p has to be equal to п i m, from which the following experimental scales could be obtained : Fig.. Pinned support P cr p = S E S l P cr m I m = I p D m = D p (7) t m = t p In this study, considering the availability of pipe size of selected model material and the loading frame capacity, the experiment was carried out in two model scales, namely (a) small model represented by seamless pipe SCH40 with the length of 600 mm and ½ inch diameter, and (b) relatively larger model that used the same pipe with the length of 000 mm and diameter of inches. Both models have the same Fig. 3. Buckling test setup for small model The load interval for small model was 1500 N (about 5% of P cr), and before buckling the interval was reduced to.5%.

4 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 14 The displacements were monitored by LVDT's at each loading stage following the load interval, and the results were used for plotting the so called P-delta diagram. B. Large model The model was derived from typical platform structure with geometric scale of 10. The total number of specimen was ten, in the form of seamless tubular pipe SCH40. The length of the models were 000 mm, and the diameter were inches, with slenderness ratio of 16. Circular holes with diameter of 0.5 pipe diameter were provided respectively at 0.15L, 0.5L, and 0.5L. ends and monotonically loaded on a loading frame by hydraulic jack untill buckling occur. The compressive load was monitored by a load cell and the loading rate was maintained at 10 kn/minute [10]. Three LVDT s were used to monitor vertical and lateral displacements at the mid high specimen and strain gauges were placed near the hole for monitoring axial strain. To prevent out of plane buckling, a pair of angle steel profiles were installed on the left and right sides of the the specimen being tested. The buckling test setup for the large model are shown in Fig. 4. For the large models, the load interval was 3000 N (about.5% of Pcr), and before buckling the interval was reduced to 1.5%. The recorded displacements at each loading stage were used for plotting P-delta diagram. C. Stress concentration measurement Stress concentration surrounding the cutout was observed on small model as well as large model, by installing two axial strain gauges, one was placed beside the cutout (compression side) while the other one was placed on the back side of the cutout (tension side) (Fig. 5). The thermal effect on strain measurement was compensated by placing dummy gauge on unloading specimen. All strain gauges had gauge factor of.1 and gauge resistance of 10 Ω, and connected by relatively low resistance copper lead wire system (0.1 Ω) to a datalogger for data acquisition. The gauge lengths used were.5 mm for small model and 6 mm for large model. (a) compression side (b) tension side Fig. 5 Strain gauges locations IV. RESULTS AND DISCUSSIONS Fig. 4. Buckling test setup for the large model Two specimens without hole were also tested for references. The pipe specimen was pinned supported at both A. Critical Buckling Loads Based on the results of small model tests (1/ inch pipe diameter and 600 mm long), relationship between the applied axial load and its corresonding lateral displacement at mid high of the specimen was plotted in Fig. 6. There are 4 P-delta curves corresponding to, respectively, specimen without any cutout and specimens with circular cutout at 0.15L, 0.5L,

5 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 15 and 0.5L. As expected, the curves show a rapid rate of change of displacements before buckling occur. The buckling modes observed corresponded with member bent away to opposite side of the cutout position. The critical buckling load (P cr) were observed to be N, 976 N, 895 N, and 733 N, respectively for specimen without cutout, specimens with cutout at 0.15L, 0.5L, and 0.5L. The reductions were 3.75%, 6.34%, and 11.53% compared to intact specimen, which were considered to be significant reduction of the critical buckling loads. Based on the Euler formula (Eq. 1), the critical buckling load of the specimen without hole for small models was 3070 N, which was closed enough to that of experiment result of N. 0.5L, respectively, where for each case represented by buckling tests. Fig. 8. P-delta curves obtained from large model tests Fig. 6. P-delta curves obtained from small model tests Fig. 7. The effect of cutout locations on small model tests Based on the results of the large model tests ( inches pipe diameter and 000 mm long), relationship between the applied axial load and its corresonding lateral displacement at mid high of the specimen was plotted in Fig. 8. The experiment consisted of 8 tests, corresponding to P-delta curves for intact specimen, specimen with circular cutout at 0.15L, 0.5L, and Similar to the previous small model test results, the curves show a rapid rate of change of displacements before buckling occur. The buckling modes observed also corresponded with member bent away to opposite side of the cutout position. From the first serie of experiment, the critical buckling load (P cr) were observed to be N, N, N, and N, respectively, for specimen without cutout, specimens with cutout at 0.15L, 0.5L, and 0.5L. The reductions were.8%, 5.38% and 1.15% respectively, compared to the intact specimen. From the second serie of experiment, the critical buckling load (P cr) were observed to be N, N, N, and N, respectively, for specimen without cutout, specimens with cutout at 0.15L, 0.5L, and 0.5L. The reductions were 7.7%, 8.07% and 11.% compared to the intact specimen. Both reductions obtained from the first and second series of tests were also agree well with those obtained from the small model tests. Based on the Euler formula (Eq. 1), the critical buckling load of the specimen without hole for large models was N, which was closed enough to that of experiment result of N. B. The effect of cutout positions To show the effect of cutout positions on the critical buckling loads, the test results from the small models were plotted as shown in Fig. 7. From the figure it was observed that : (a) P cr tended to decrease when the shortest distance between cutout locations and member end increased, and (b) the maximum reduction occurs when the cutout position was in the middle of the member length. The reduction of P cr were also calculated from Equation (4). As shown in Fig. 7, the trend was similar to this research

6 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 16 results, however, the differences in the intensites of P cr were relatively large. small models and on large models. This was observed by measuring the strain near the hole. Through strain gauges mounted next to the hole and at the opposite side of the pipe (side without holes) (Fig.5) the strain intensity and associated strain direction that developed on the pipe near the hole could be observed. The strain gauges were mounted on both models, with hole positions at 0.15L, 0.5L and 0.5L. For the small models, the observed relationship between the applied load and measured strain on the hole side are presented in Fig. 11, while that on the opposite side are presented in Fig. 1. For the large models, load-strain observations on hole sides are presented in Fig. 13, and that on the reverse side are presented in Fig. 14. Fig. 9. The effect of cutout locations on large model tests The effect of cutout positions on the critical buckling loads obtained from the large models were plotted in Fig. 9. From the figure it was also observed that : (a) P cr tended to decrease when the shortest distance between cutout locations and member end increased, and (b) the maximum reduction occurs when the cutout position was in the middle of the member length. The first and the second series of experiments were distinguished by the slightly different quality of material due to different sample purchasing time. Fig. 11. Load-strain curves at the hole side of the small model Fig. 1. Load-strain curves at opposite hole side of the small model Fig. 10. The effect of cutout positions on Pcr Very good agreement between the results of small and large models were shown in Fig. 10, by normalizing P cr and cutout posisiton L 0 became nondimensional variables P cr/p * cr and L 0/L, where P * cr denoted the critical buckling load of the member without cutout, and L 0/L is the relative position of the cutout. The observed strain development during both small and large model tests showed that initially when the load was relatively small (up to about 5% Pcr), the two strain gauges on the side of the hole and on the opposite side of the hole showed compressive strain. But as the load increased, the strain on the side of the hole remain compressive and gradually enlarges until buckling occurs, while the strain on the opposite side gradually changed into tensile strain, and it continued to increase until buckling occurs, as shown in Fig. 11 and 1 for small models, and in Fig. 13 and 14 for large models. C. Strain around the cutout The existence of cutout on pipe under compressive axial load will create stress concentration around the holes, both on

7 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: 0 17 Fig. 13. Load-strain curves at the hole side of the large model Fig. 14. Load-strain curves at opposite hole side of the large model Both on the compressive and tensile sides, the maximum strains before buckling occur were ranging from 0.06%, to 0.1%. The observed strain values were lower than the steel proportional strain limit of about 0.15%. This is consistent with the basic theory of elastic buckling that the buckling in this experiment was related to elastic buckling phenomenon or geometric nonlinearity, caused by a slenderness ratio of the pipe exceeding 100. Therefore the stress on the pipes were still far below the yielding stress of the steel material. D. Buckling mode and hole deformation during buckling Observations from the 1 buckling modes of the pipe specimens with cutout presence, both on small and large models for various hole positions, indicated that the buckling mode corresponded with member bent away to opposite side of the cutout position (Fig 15). This is consistent with the fact that on the hole side there will be a stress concentration, i.e.: the intensity of the stress is relatively greater than the average stress, resulting in greater deformation and in turn initiating the buckling process with the shape of buckling mode corresponded with member bent away to opposite side of the cutout position. A typical form of cutout deformation when buckling occurs are presented in Fig. 15. Fig. 15. Buckling mode and hole deformation during buckling It should be noted that all experimental results presented herein, i.e.: the critical buckling loads and their associated P- delta curves, the buckling modes, hole deformation, and the effects of cutout positions on the critical loads, had been verified numerically by finite element method involving buckling analyses as well as nonlinear geometric analyses. The numerical results are presented in another paper. V. CONCLUSIONS Based on the experimental test results on specimens in the form of steel tubular compressive members with circular cutout presence, the following conclusions could be drawn: 1. The presence of cutout reduced the critical buckling load of slender tubular compressive member significantly.. The reductions ranging from.8% to 1.15%, depending on the cutout posistions, 3. The maximum reduction occurs when the hole position was in the middle of the member length. 4. Initially when the load was relatively small (<5% Pcr), both sides of the pipe experienced compressive strain, however as the load increased, the strain on the side of the hole remain compressive, while the strain on the opposite side gradually changed into tensile strain, and both continued to increase until buckling occurs.

8 International Journal of Civil & Environmental Engineering IJCEE-IJENS Vol: 17 No: The buckling modes corresponded with member bent away to opposite side of the cutout position. ACKNOWLEDGMENT The authors would like to thank to the Structural Engineering Laboratory, Department of Civil and Environment Engineering, Faculty of Engineering, Gadjah Mada University, for providing continuous and inspirative support in conducting this experiments. REFERENCES [1] Iman, M., Model Kegagalan Tekuk Euler Struktur Kolom CHS Dengan Pengaruh Pitting Korosi Pendekatan Metode Elemen Hingga (SAP 000 Versi 11,00) Dan Uji Laboratorium. Master Theses, Graduate Program, Gadjah Mada University, 010. [] American Institute of Steel Construction, Specification for Structural Steel Buildings. ANSI/AISC , 010. [3] Haipeng, H., Jinquan, C., Farid T., and Neil, P., Numerical and experimental investigations of the response of alumunium cylinders with a cutoutout subject to axial compression. Thin Walled Structures, Vol.44, 54-70, 006. [4] Shariati, M., and Masoud, M.R., Numerical and experimental investigations on buckling of steel cylindrical shells with elliptical cutout subjected to axial compression, Thin Walled Structure. Vol 46, , 008. [5] Shariati, M., Fereidoon, A., and Akbarpour A., Investigation on buckling behavior of tubular shells with circular cutout, subjected to combined loading, Research-Journal of Recent Sciences. Vol.1 (7), 68-76, 01. [6] ASTM, Standard Guide for Examination and Evaluation of Pitting Corrosion, [7] Chen, W.F. and Atsuta, T., Theory of Beam Columns, In-plane Behavior and Design, Vol. 1, McGraw-Hill Inc., [8] Samodra, M.P.. Off shore platform reappraisal by using SAP000 V.11 software: static, seismisc and fatigue analyses. Master Theses, Graduate Program, Gadjah Mada University, 008. [9] AISI, Specification for the Design of Cold~Formed Steel Structural Members, [10] Ghazijahani, T.G., Jiao, H. & Holloway, D. Structural behavior of shells with different cutouts under compression: An experimental study. Journal of Constructional Steel Research. 105 (015) [11] Ghazijahani, T.G., Jiao, H. & Holloway, D., Plastic Buckling of Dented Steel Circular Tubes Under Axial Compressiion : An Experimental Study. Thin Walled Structures, 0(015)