J. Construct. Steel Research 34 (1995) 27-52 © 1995 Elsevier Science Limited Printed in Malta. All rights reserved 0143-974X/95/$9.50 Tests of High Strength Steel Columns K. J. R. R a s m u s s e n & G. J. H a n c o c k School of Civil and Mining Engineering, University of Sydney, New South Wales 2006, Australia (Received for publication 25 January 1994) ABSTRACT The paper describes a test programme on columns fabricated from high strength steel plates with nominal yield stress of 690 M P a. The programme comprised 13 box and I-section specimens, including fixed-ended stub columns and pin-ended long columns. For the pin-ended columns, two tests were performed for each length using eccentric and concentric axial loading. The purpose of the test programme was to select a curve for high strength steel columns with a nominal yield stress of 690 MPa from the multiple column curves used in the Australian steel structures standard. It is shown that the ctb= -0.5 curve is the appropriate curve for box and I-section columns fabricated from flame-cut high strength steel plate with a nominal yield stress of 690 MPa. This curve is higher than the otb = 0 curve for box and I-section columns fabricated from ordinary steel because the effect of residual stresses is less detrimental to the strength of high strength steel columns than to the strength of ordinary steel columns. The paper also shows a comparison of the tests with column design strengths of the Australian steel structures standard AS4100, the load and Resistance Factor Design Specification of the American Institute of Steel Construction, The British Standc~rd BS5950: Part l, and the draft European Committee for Standardisation (CEN) Eurocode3. The design strengths are shown to be in close agreement with the tests except for Eurocode3 which conservatively predicts the test strengths. NOTATION A Ae A~ An b bf bw Cross-section area Effective cross-section area (AS4100) Gross cross-section area (AS4100) Net cross-section area (AS4100) Plate width Clear flange width of I-section Web width of I-section 27 28 Bf e eo E fE I I kf L Le Lt Nc Ns Py Pu F t tf tw tw V VO V 0~a ~b ~c /3 u r/ 2,2. O" O'rf O'rw O"u O'yc ~ O'yt K. J. R. Rasmussen, G. J. Hancock Total width of I-section flange Total deviation of the centroid at mid-length from the line of action of the force Measured eccentricity of applied load Young's modulus Critical flexural buckling stress Yield stress Measured current during welding Second moment of area about buckling axis Form factor (AS4100) Test specimen length Effective column length Total pin-ended column length Column strength (AS4100) Section capacity (AS4100) Design strength (BS5950) Ultimate load Radius of gyration about buckling axis Plate thickness Thickness of flange plate of I-section Thickness of web plate of 1-section Leg length of fillet weld Measured velocity of welding arc Measured overall geometric imperfection at mid-length Measured voltage during welding Constant defining the imperfection parameter Modified column slenderness (AS4100) Constant defining multiple column curves of AS4100 Slenderness reduction factor Average strain in coupon tests Strain corresponding to ultimate tensile stress Imperfection parameter Column slenderness measures (AS4100) Column slenderness 4fy/fl~ Constant (AS4100) Average stress in coupon tests Average residual stress in flanges of 1-section Average residual stress in web of I-section Ultimate tensile strength Compressive/tensile yield stress Parameter in Perry formula Tests of hiah strenath steel columns 29 1 INTRODUCTION The scope of the current Australian Standard 1 for steel structures is limited to ordinary steel with yield stress less than 450 MPa. Consequently, in Australia, structural members fabricated from high strength steel (defined in this paper as steel with a yield stress in excess of 450 MPa) are usually designed according to overseas specifications which allow the use of high strength steel, notably the American Institute of Steel Construction (AISC) Load and Resistance Factor Design (LRFD) Specification. 2 A similar situation exists in Britain and Europe. Section 3.1.1 of the British Standard BS5950: Part 13 specifies that 'the standard covers the design of structures fabricated from steels supplied to BS4360;* other steels may be used provided due allowance is made for variations in properties, including ductility'. Hence, although high strength steels are not ruled out explicitly, reference is given to BS4360 which concerns ordinary steels only. Also, throughout the standard, the yield stresses quoted are those of ordinary steels. The 1983 draft of the Eurocode3 specification s contained separate rules for high strength steels, including design curves for I-section columns fabricated from high strength steels that were higher than those for I-section columns fabricated from ordinary steels. These rules were not included in the 1992 draft of the specification which gave reference to separate rules for high strength steels, contained in Annex D of the specification. However, this annex applies specifically to hot-rolled Isections with nominal yield stress less than or equal to 460 MPa. The tests described in this paper form part of a research programme into the strength of members fabricated from high strength steel. The aim of the programme is to investigate whether high strength steel members with yield stresses in the range from 450 MPa to 700 MPa can be designed according to existing rules of the Australian steel structures standard (AS4100) or whether these rules need to be modified to include high strength steel. According to the rules of AS4100, the column strength is determined as a product: of the section capacity (stub column strength) and a slenderness reduction factor which accounts for the reduction in strength resulting from overall instability. The Australian Standard uses multiple column curves, such that the appropriate slenderness reduction factor depends on the type of cross-section and manufacturing process. In Ref. 6, tests were described on stub columns fabricated from high strength steel with a nominal yield stress of 690 MPa. These tests concerned the section capacity and were performed to investigate whether the yield slenderness limits for welded uniformly compressed plates supported 30 K. J. R. Rasmussen, G. J. Hancock along one or both longitudinal edges were applicable to high strength steel plates. The yield slenderness limit was defined as the width to thickness ratio (factored by x//~-y/250 where fy was the yield stress in MPa) beyond which the ultimate load was reduced below the squash load as a result of local buckling. In this paper, tests are presented on long columns fabricated from high strength steel with a nominal yield stress of 690 MPa. The purpose of the tests was to determine the appropriate slenderness reduction factor to use for columns fabricated from high strength steel with a nominal yield stress of 690 MPa. The tests were performed on box and I-section column fabricated by welding. It was established from analytic modelling7 and testings that the strength of columns fabricated from high strength steel exceeded the strength of columns of equal length and cross-section fabricated from ordinary steel when compared on a nondimensional basis. The difference arose because the ratios of residual stress at critical points in the cross-section to the yield stress were less for high strength steel columns than for ordinary steel columns, and it was shown that it was the ratio of residual stress to the yield stress, rather than the magnitude of residual stress itself, which governed the reduction in strength. The ratio was lower for high strength steel because the magnitude of residual stress was largely independent of the yield stress. 9 Consequently, it should be expected that it might be possible to use a higher value of the slenderness reduction factor for columns fabricated from high strength steel than the value adopted in AS4100 for columns fabricated from ordinary steel, so that ~he influence of residual stress on the strength of high strength steel columns is reduced in design. Nishino and Tall s performed tests on columns rolled or fabricated from high strength ASTM 514 steel with a nominal yield stress of 690 MPa. In these tests, the columns were loaded 'concentrically' in accordance with the American design philosophy which was to base the design strength on straight concentrically loaded columns and to allow for imperfections by using a comparatively high factor of safety. (The positions of the loading points were adjusted such that the effects of geometric imperfections and loading eccentricity counteracted approximately each other, and hence the columns failed by nearly perfect bifurcation buckling.) Consequently, these tests could not be used to select a column curve for high strength steel columns in the Australian Standard, since the rules of this standard are based on the strength of columns with overall geometric imperfections of one-thousandth of the length. Tests of high strength steel columns 31 The purpose of this paper is: • • • to present a test program on long box and I-section columns fabricated from high strength steel with a nominal yied stress of 690 MPa; to show the selection of a strength curve to be used in the Australian Standard for columns fabricated from high strength steel with nominal yield stress of 690 MPa; to present a comparison of the test strengths with design strengths obtained using the AS4100, AISC-LRFD, BS5950: Part 1, and draft Eurocode3 specifications. 2 TEST PROGRAMME 2.1 OutLine of test programme The test programme comprised 13 box and I-section columns tested as fixed-ended stub columns and pin-ended long columns. For each type of cross-section, the lengths of the pin-ended tests produced column slenderness values (L/r) or about 30, 50 and 90, and hence normalised slenderness value (~[= ~ y / f : ) of about 0.55, 0-95 and 1.70. For each pin-ended length, two tests were performed. In the first, the load was applied approximately at the geometric centroid while in the second, i~t was applied with a nominal eccentricity of one-thousandth of the lengt]~. The second test was performed to determine the strengths of columns with initial out-of-straightness, the allowable limit of which is one-thou:~andth of the length according to AS4100. Since the initial out-of-straightness of the test specimens was generally much less than a thousandth of the length, a loading eccentricity of this magnitude was introduced as an equivalent geometric imperfection. The test results of the longest concentrically loaded I-section were considered unreliable and have not been reported in this paper. One stub column was tested between fixed ends for each type of cross-section. The stub column lengths complied with the recommendations of Ref. 10, and so were sufficiently short to exclude overall instability effects, yet sufficiently long to allow unrestrained development of local buckles and to eliminate effects of frictional restraints at the ends on the plate strength. Thus, the stub column strengths could be assumed to be accurate measures of the compressive section capacities. 32 K. J. R. Rasmussen, G. J. Hancock The slenderness values of the plates comprising the box and I crosssections were chosen slightly lower than the yield slenderness limits such that the sections should be able to support their squash loads. Separate specimens of each section type were prepared to determine welding residual stresses, as described in Section 2.4. Tension and compression coupons were tested to determine the mechanical properties of the steel, including Young's modulus, the tensile and compressive yield stress values, and the ultimate tensile strength, as described in Section 2.5. The test specimens have been labelled so that the cross-section and the type of test can be identified from the label. The first letter signifies the cross-section and is 'B' and T for the box and I-sections respectively. For the long column test specimens, the numbers following the cross-section type are the nominal pin-ended length in millimetres, and the letter following signifies whether the load was applied concentrically (C) or eccentrically (E). If the specimen was tested as a stub column or used for residual stress measurement, the section type is followed by 'SC' or 'RS' respectively rather than the column length. 2.2 Test specimen data and fabrication procedure The test specimens were fabricated by manual Gas Metal Arc Welding (GMAW) from nominal 5 and 8 mm plates of BISALLOY80 steel with nominal yield stresses of 650 and 690 MPa respectively. BISALLOY80 is an Australian-produced quenched and tempered steel complying with the mechanical property requirements of ASTM A514 steel. The 5 mm plates were guillotined into strips while the 8 mm plates were flame-cut. The strips were tack-welded into sections before final welding. A preheat of 50 °C was used for the 8 mm plates while the 5 mm plates were welded at ambient temperature. A single run of weld was laid along each fillet so that four runs were laid for each section type, as shown in Fig. 1. To reduce weld shrinkage deformations, the welds were laid alternately at the four fillets, rather than continuously along each fillet, and were staggered (or stepped-back) such that the staggered weld length was 400--600 mm The welding parameters used in the fabrication of the test specimens are shown in Table 1. In the table, I and V are the current and voltage displayed on the welding machine respectively, and v is the observed velocity of the weld arc. The electrode was a Lincolnweld L50 wire with a minimum nominal yield stress of 360 MPa. The nominal and measured cross-section dimensions of each column are shown in Tables 2 and 3 for box and I-sections respectively. The symbols used in these tables are defined in Fig. 1. The measurements are based on average values of width and thickness of each component plate of the Tests of high strength steel columns a_ 33 Bf _¢ r- I -I I b b •, ~ - - - bf t I Weld a) Box-section b) I-section Fig. 1. Definition of symbols. Table 1 Welding Parameters Section Box section I-section I V v (amp) (volt) (mm/s) 260 240 22 22 8.1 3.8 cross-section. The tables also include the area (A), minor axis second moment of area (I), and radius of gyration (r = ~ / ~ ) . The measured specimen lengths (L) and the pin-ended column lengths (Lt) are also shown in Tables 2 and 3. The pin-ended lengths were obtained as the sum of the specimen length and the total length (450 mm) of the end bearings. (At each end, the length from the axis of rotation to the face of the end platen, on which the specimen bore, was 225 mm.) The ends of the test specimens were milled fiat to within 0.005 mm before testing to allow proper seating on the rigid end platens of the testing rig:, and strain-gauges were attached to the pin-ended specimens at mid-lengtih to measure longitudinal strains. For the box columns, one gauge wa~ attached at the centre of each component plate, while for the I-sections two gauges were attached at 5 mm from each free edge on either side of the flange such that eight gauges in total were attached to each I-section column. K. J. R. Rasmussen, G. J. Hancock 34 Table 2 Measured Dimensions and Ultimate Loads of Box Section Test Specimens Specimen b t L Lt A I r P, (ram) (ram) (mm) (mm) (mm2) (ram4) (mm) (kN) Nominal section 90 5 -- -- 1850 2.75 x 106 38.6 -- Bll50C B1150E B1950C B1950E B3450C B3450E 88"9 87'6 88"3 89"4 90"2 89"9 5"00 4"95 4"96 4"97 4'97 4"94 699 700 1500 1500 3001 3001 1149 1150 1950 1950 3451 3451 1828 1783 1801 1827 1843 1825 2"64 x 106 2"51 x 106 2"57 x 106 2'67 x 106 2"74 x 106 2"69 x 106 38"0 37-5 37"8 38"2 38"6 38"4 1174 1137 1078 926 469 438 BSC 88"8 4"95 400 400 1807 2"61 x 106 38"0 1236 2.3 Column tests 2.3.1 Geometric imperfections Overall geometric imperfections are here defined as the deviation of the column axis at mid-length from a straight line connecting the ends, and are denoted by Vo, as shown in Fig. 2. The imperfections of the box columns were measured in both principal planes, whereas for the 1-sections only minor axis imperfections were measured. During measurement, the ends were simply supported and the specimen allowed to sag between the supports. Readings were taken using the optical level at the ends and at the centre, allowing the deviation of the column axis at mid-length from a straight line connecting the ends to be calculated. This procedure was repeated after rotating the column by 180 ° a b o u t its longitudinal axis, and the two readings were averaged to eliminate gravity effects. The imperfections (v0) are shown in Tables 4 and 5 for box and I-sections respectively. F o r the box sections, the imperfection is the largest of the two measured values for each specimen. Only the largest imperfection has been included in Table 4 because all box columns failed by bending in the principal plane that included the largest geometric imperfection. bf tf bw tw tw L 140 141"5 141-1 141"5 141"5 140"3 140"0 I1000C I1000E I1650C I1650E I2950E ISC 66' 1 66"9 66"7 66"9 66"9 66'3 66 7'73 7"70 7"67 7"70 7"71 7"75 8 8 1 4 2 " 0 7"73 1 4 0 - 0 7"70 141"8 7"71 1 4 1 - 5 7"66 1 4 3 " 0 7"75 1 4 2 " 0 7"74 140 6"0 6"0 6'8 5"1 5"3 6"2 6 400 550 550 1199 1199 2500 -- (ram) (ram) (ram) (ram) (ram) (ram) (ram) Bf Nominal section Specimen 400 1000 1000 1649 1649 2950 -- (mm) Lt 3334 3329 3350 3315 3346 3351 3430 (ram2) A 106 106 106 106 10 6 10 6 3"54 × 106 3"64 X 3"60 x 3"64x 3"65 x 3"57 x 3.66 × (aim4) I Table 3 Measured Dimensions and Ultimate Loads of I-Section Test Specimens 32'6 33"1 32"8 33"1 33"0 32-6 32"7 (ram) r 2369 2092 2192 1751 1682 745 -- (kN) Pu t~ t~ K. J. R. Rasmussen, G. J. Hancock 36 / ~ ~ " -- ~--~. _~ Fig. 2. Loading eccentricity and geometric imperfection. Table 4 Nondimensional Test Strengths of Box Section Columns Specimen Vo eo e/Lt × 103 Le/r 2n = Le/r x/~r~/250 Pu/(A~rv,) (ram) (ram) B1150C B1150E B1950C B1950E B3450C B3450E 0"4 1"2 0"6 1"0 1"0 3"8 0-1 0"9 -0-1 2"2 -0"6 -0"9 0"43 1'83 0-26 1"64 ~12 0'84 30"2 30'7 51"6 51"0 89-4 89"9 50"7 51"6 86-7 85'6 150"1 151"0 0"911 0'905 0"849 0"719 0"361 0'340 BSC -- -- -- 5"3 8.9 0"970 Table 5 Nondimensionai Test Strengths of I-Section Columns Specimen Vo eo e/Lt x 10a Le/r 2n=L©/r x/~vJ250 Pu/(Aav~) (ram) (ram) I1000C I1000E I1650C I1650E I2950E 0'0 0"1 0"3 0-5 0'6 0"7 1.2 0"1 0"5 1.4 0"70 1"30 0"25 0"61 0"68 30"2 30-5 49"8 50"0 90"5 49.1 49.5 81"0 81"2 147.0 0"952 0-991 0"800 0"762 0"337 ISC -- -- -- 6"1 10"0 1-077 2.3.2 Loading eccentricity T h e e c c e n t r i c i t y (Co) o f t h e a p p l i e d l o a d at t h e s u p p o r t s , as s h o w n in Fig. 2, w a s c a l c u l a t e d for e a c h l o n g c o l u m n d u r i n g initial l o a d i n g b y m e a s u r i n g t h e d e f l e c t i o n a n d l o n g i t u d i n a l s t r a i n s at m i d - l e n g t h . T h e e c c e n t r i c i t y (eo) Tests of high strength steel columns 37 and the total deviation (e=vo +eo) of the centroid at mid-length divided by the pin-ended length (Lt) are shown in Tables 4 and 5 for each test specimen. The deviation (e) was measured from a straight line connecting the points of application of the force at the ends, as shown in Fig. 2. In the tests of the long columns, the specimens were positioned in the rig such that e/Lt x 10a was approximately zero and unity for the concentrically and eccentrically loaded columns respectively. As shown in Tables 4 and 5, the measured values of e/Lt x 103 differ slightly from these nominal values, reflecting the difficulty of positioning the specimens accurately in the rig. 2.3.3 Te.~t procedure The long columns were tested between pinned end-bearings in a horizontal serve-controlled Dartec test rig. The stub columns (BSC and ISC) were tested between fixed ends. The instrumentation of the pin-ended columns consisted of a load cell measuring the axial force, transducers measuring axial compression, deflections in the principal directions at mid-length, and end rotations, as well as strain-gauges measuring longitudinal strains at mid-length. After exceeding the proportionality stress of the material, readings were taken approximately 1 rain after applying an increment of shortening to allow the stres,; relaxation associated with the mobilisation and locking of dislocatie,ns between metal crystals to take place, tl The ultimate loads (Pu) of each long column and each stub column are shown in Tables 2 and 3. 2.4 Residaai strains The longitudinal membrane residual strains of the box and I-section were obtained from specimens (BRS and IRS) with the same nominal crosssections its the long columns and stub columns. The residual stress specimens were fabricated from the same virgin plates using the same welding procedure as used for the long columns and stub columns, as described in Section 2.2. The strains were obtained using the sectioning technique. For the I-section, readings were taken on each side of the component plates using an extensometer before and after slicing the section into narrow strips to release the residual strain. Readings on opposite sides were averaged to obtain the membrane strain which was subsequently converted to a membrane; stress by multiplying the strain by Young's modulus (E=210GPa), as obtained from the compression coupon tests. The residual stresses measured in the 1-section specimen are shown in Fig. 3. K, J. R. Rasmussen, G. J. Hancock 38 -200 s 1 -100 0 100 200 MPa qL (, ~l 15m /, • lOmm MPa 200 100 0 -100 -200 % = -135HPa or, = -32HPa Fig. 3. R e s i d u a l stress m e a s u r e m e n t for I-section. The figure also shows the locations of the residual stress measurements in the cross-section and the average of the compressive residual stress measurements in the web (~,,) and in the flanges (~rf), assumed to be positive as tensile. The calculation of the average compressive residual stress in the flanges (~rf) did not include the measurements closest to the free edges of the flanges, since these measurements were tensile because the virgin plates were flame-cut into strips. The average compressive residual stress is a useful measure of the level of residual stress which may have a weakening effect on a column. The residual strains of the box sections were measured using strain gauges. Gauges were attached on both sides of each component plate near the half-width. Readings on opposite sides were averaged to provide the membrane strain which was converted to a membrane residual stress using the measured compressive Young's modulus. The residual stress measurements and their average (~r,= - 123 MPa) are shown in Table 6. 2.5 Tension and compression coupon tests Tension and compression coupons were cut from the same 5 and 8 m m virgin plates from which the long column, stub column and residual stress specimens were manufactured. The coupons were cut parallel to the rolling Tests of high strenoth steel columns 39 Table 6 Residual Stress of Box Columns Residual stress (MPa) Side 1 - 145 Side 2 - 92 Side 3 - 148 Side 4 - 105 Average - 123 direction and equipped with two strain gauges on opposing sides at mid-length to measure longitudinal strains for accurate determination of Young's modulus. The compression coupons were tested in a jig preventing lateral buckling. The tension and compression coupons were tested in accordance with AS1391 t2 using a low strain rate (< 15 #e/s) at strains less than 20 000 #5. At strains greater than 20 000 #e, the strain rate of the tensile coupons was increased to approximately 500#e/s. The ultimate tensile strength (o-u) and the corresponding strain (eu) were measured near the conclusion of the test after pausing the applied straining for 1 min. The stress-strain curves obtained from the tension and compression coupon tests are shown in Fig. 4(a) and (b) for the 5 and 8 mm plates respectively. In the figures, the strain (5) is the average of the two strain gauge readings and the stress (tr) is the measured load divided by the initial area calc,ulated using the coupon dimensions measured before testing. For strains greater than 20000#8, the strain of the tension coupons was obtained by dividing the cross-head displacement by the initial parallel length. As shown in Fig. 4(a) and (b), the scale on the strain axis changes at 20 000/~e. This was also the strain at which the strain rate was increased in the tests. The mechanical properties obtained from the coupon tests are summarised Jin Table 7. In the table, Young's modulus (E) was obtained using a linear regression analysis of the stress-strain points for stresses up to 300MPa. The compressive (trvc) and tensile (trvt) yield stresses were obtained as the 0.2% proof stresses, as shown in Fig. 4(a) and (b). Table 7 also contains the nominal values of tensile yield stress, ultimate tensile strength and equivalent percentage elongation based on a gauge length of 50 mm. 40 K. d. R. Rasmussen, G. J. Hancock 900 aye = 7kSMPa / 800 0 (MPa) Compression coupon 700 / Te n : i : : S c ~ ; o n 600 50O J r !, Strain rate: SOOpc/s t,O0 300 200 100 0 f" re Strain rate: 1Spe/s // / I 0.2 1 Change of scale eu= "/.6% " i . I I 2 I I I /, ¢ I 6 10z x I I 8 I I 10 I I 12 I I 16 1~ (a) 900 aye = 660MPa Compression coupon 800 0 (MPa) 700 600 500 tl 7 ~ O y t = 660MPa %= 725MPa Tension coupon / L,O0 Strain rate: SOOp¢/s Fraci'ure Str~n rate: 15pe/s 3O0 200 I 0 eu= 8./.% Changeof scale IO0 I 0 0.2 . I " I i 2 4 I ~: x I t 6 t i 8 i i 10 i i 12 t i 1/, 102 (b) Fig. 4. Stress-strain curves for (a) a 5 mm plate and (b) an 8 mm plate. i 16 Table 7 Tension Compression Tension Compression 5 5 8 8 690 650 tryt (MPa) ° Based o n 50 m m gauge length. Compression or tension test Nominal plate thickness 790-930 750-900 (MPa) o. 18 18 (%) Percentage elongationa Nominal values 208 213 214 213 (GPa) E 725 -- -660 660 750 o"u 745 -- (MPa) Oyc -- 705 6"yt Measured values N o m i n a l a n d M e a s u r e d M e c h a n i c a l Properties of 5 a n d 8 m m Plates 10 10 (%) Eu 42 K. J. R. Rasmussen, G. J. Hancock 3 COLUMN CURVE SELECTION 3.1 Column design rules of AS4100 According to Section 6 of AS4100,1 the capacity of a column is obtained as (1) N<=~¢N, where ~c is the member slenderness reduction factor and N, is the section capacity. 3.1.1 Section capacity The section capacity (N,) is calculated a~ (2) Ns=krAnfy where A, is the net cross-section area, fy the yield stress, and kf the form factor, given by kf=~ (3) In eqn (3), A e and A s are the effective and gross cross-section areas respectively. It was recommended in Ref. 6 that the yield slenderness limits specified in AS4100 should also be used for high strength steel plates. The yield slenderness limit was defined as the width to thickness ratio (factored by x/fy/250 where fy was the yield stress in MPa) beyond which the ultimate load was reduced below the squash load as a result of local buckling, In the present paper, the slenderness values of the plates comprising the test sections were chosen slightly lower than the yield slenderness limits, and so the net and gross areas were the same and equal to the cross-section area (A) shown in Tables 2 and 3. Consequently, the section capacity (Ns) was equal to the squash load (,4 fy) for both cross-sections. 3.1.2 Slenderness reduction factor The slenderness reduction factor (0t~) accounts for the reduction in capacity resulting from overall instability. It is calculated as ] (4) Tests of high strength steel columns 43 where +1+7 (5) 2=2,+=a~b (6) r/=0-£03 26(4-13-5)>10 (7) (8) 2100(2.- J3.5) 2gl- 15"34, + 2050 (9) In eqn (8), L, and r are the effective length and radius of gyration respectively, and in eqn (6), the member section constant (~tb) attains one of the values, -1.0, -0.5, 0, 0.5, 1.0, depending on the method of manufacture, type of cross-section and thickness of component plates. The five values of ~b define the multiple column curves used in AS4100, as shown in Fig. 5. For welded I-sections fabricated from flame-cut plates and for welded box sections, the column curve specified in AS4100 is the ~b = 0 curve. 3.2 Comparison of test strengths with design strengths according to AS4100 The column slenderness (2. = Le/rx/~vJ250) and nondimensional strength (P,/(Atrv,)) of the box and I-sections are shown in Tables 4 and 5 respectiw;ly. In determining 2 , and Pu/(Atrv,), Le was equal to the pin-endecl column length (Lt), r and A were calculated using measured values of length and cross-section dimension, as shown in Tables 2 and 3, and try, was the measured tensile yield stress shown in Table 7. The nondimensional test strengths of the box and 1-sections are compared with the five column curves of AS4100 in Fig. 5(a) and (b) respectively. As shown in Fig. 5(a), the nondimensional strength (P,/(Aav,)=0"970) of the box section stub column is 3% lower than the squash load. This discrepancy is a result of the plate strength formulae adopted :in AS4100 which produce optimistic section capacities when 44 K. J. R. Rasmussen, G. J. Hancock 1.4 I I I I I I I I I / 1.2 1.0 e~ 0.8 0.6 OA ab 0.5 0.2 0 (a) o • Concentric loading Eccentric loading 0 I I I I I I I I l 20 40 60 80 100 120 140 160 180 200 1.4 12 Tt 10 , 0.8 -- ~ ~ , / Pu Ao~t Euler 0.6 -- -1. -0.5 ° loading • 0 Eccentric loading I I I I I h I I I 20 40 60 80 100 120 140 160 180 200 (b) Fig. 5. C o l u m n curves o f A S 4 1 0 0 a n d test strengths for (a) b o x c o l u m n s a n d (b) I-section columns. applied to high strength steel plates supported along both longitudinal edges with slenderness values near the yield slenderness limit, as discussed in detail in Ref. 6. The discrepancy is not related to the choice of column curve, since the five curves merge at short column lengths. It should be noticed that the lack of conservatism would have been less if the c o m p o n e n t plates had Tests of high strength steel columns 45 been either stockier or more slender, since the plate slenderness value of the BSC section was close to the yield slenderness limit, which was the slenderness value for which the standard produced the most optimistic plate strengths. The strengths of the shortest pin-ended box columns (Bll50C and Bll50E) are approximately equal, despite the larger loading eccentricity applied in the test of Bll50E compared to the test of Bl150C. This result may be explained by the fact that short columns are fairly insensitive to eccentric: loading. The strength of the nominally concentrically loaded box column of intermediate length (B1950C) is 18 % higher than the strength of the eccentrically loaded specimen (B1950E) of equal length, as shown in Fig. 5(a). The difference is primarily attributed to the larger eccentricity used in the test of B1950E compared to that used in the test of B1950C. It should be noticed that the eccentricity (e = 1.64 Lt/1000 ) used in the test of B1950E exceeded the nominal value of Lt/lO00. The strength of the longest concentrically loaded box column (B3450C) is nearly equal to the Euler load, as shown in Fig. 5(a). In the test, the column remained nearly straight until reaching the ultimate load, at which the mid-length deflection increased rapidly, exhibiting nearly perfect bifurcation behaviour. As a result of the eccentricity incorporated in the test of B3450E, the strength of this column is 6% lower than the concentr~ically loaded column (B3450C). However, the eccentricity (e=0.84Lt/1000) used in the test of B3450E was less than the nominal value (Lt/1000). In sele,cting a strength curve for box columns fabricated from high strength steel, the test strengths of the eccentrically loaded columns shall be compared with the multiple column curves shown in Fig. 5(a). The ~b = --0"5 curve is in closest agreement with the strengths of the test sections of intermediate lengths. The shortest pin-ended column strength (Bll50E) is slightly lower than the ~b = - 0 " 5 curve because at short lengths the strength approaches the stub column strength which is less than the squash load for reasons explained previously. Considering the full range of lengths, the ~b = --0"5 curve appears to be the appropriate curve for box columns fabricated from high strength steel with nominal yield stress of 690 MPa. As shovcn in Fig. 5(b), the strength (Pu/(Atrvt)= 1'077) of the I-section stub column (ISC) is 8% higher than the squash load. The strength of the shortest concentrically loaded pin-ended I-section (I1000C) is less than the strength of the eccentrically loaded I-section (I1000E) of equal length, as shown in Fig. 5(b), despite the larger loading eccentricity used in the test of I1000E. This result may be explained by the fact that the strengths were K. J. R. Rasmussen, G. J. Hancock 46 influenced by random variations of residual stresses and geometric plate imperfections and that these variations may have had a greater influence on the strengths than the loading eccentricity. The strengths of the eccentrically loaded I-section specimens of intermediate and long length are slightly higher than the 0%= -0"5 curve, as shown in Fig. 5(b). However, the loading eccentricity used in these tests was slightly smaller than Lt/lO00, as shown in Table 5. As for box columns, the ~b = --0"5 curve appears to be the appropriate column curve for I-sections (minor axis buckling) fabricated from flame-cut high strength steel plate. On the basis of the comparison shown in Fig. 5(a) and (b), it has been recommended that the ~b = --0"5 curve be used in the Australian Standard for welded box columns and welded I-section columns (minor axis buckling) fabricated from flame-cut high strength steel plate (t < 40 mm) with a nominal yield stress of 690 MPa. 4 C O M P A R I S O N O F TEST S T R E N G T H S W I T H T H E A U S T R A L I A N , A M E R I C A N , BRITISH A N D E U R O P E A N S P E C I F I C A T I O N S F O R STEEL S T R U C T U R E S 4.1 General The test results of the box and I-section columns are compared in Figs 6 and 7 with design strengths obtained using the Australian, 1 American,2 British 3 and European 5 specifications for steel structures. The figures also include the measured nondimensional eccentricities (e/Lt x 103), as obtained from the eccentrically loaded tests. The comparison is based on the nominal cross-sections detailed in Tables 2 and 3, and a nominal yield stress of 690 MPa. The Eurocode3 design curves shown in Figs 6 and 7 are based on Section 5.5.1 of the specification. (Annex D of Eurocode3 allows a higher design curve to be used for I-sections of nominal 420 and 460 MPa yield stress than sections of ordinary European steel, having nominal yield stresses of 225, 275 and 355 MPa. However, this annex applies specifically to hot-rolled sections.) For each specification, the component plates of the cross-sections were sufficiently stocky that the section capacity was equal to the squash load (Afy). However, in using the British Standard, the design strength (py) was reduced by 20 MPa in accordance with Section 4.7.5 of that standard because the columns were fabricated by welding. Tests of high strength steel columns 47 \1 1.500 / 1,250 1,000 750 500 250 _ - - ..... 0 0 AISC - LRFD BS5950: Part 1 Eurocode 3 I I 20 40 I 60 o Concentric loading * Eccentric loading I 80 100 L/r Fig. 6. C o m p a r i s o n of design s t r e n g t h s a n d test s t r e n g t h s for b o x c o l u m n s using n o m i n a l values. 3,000 [ I \ [ 1,01111 I "L~/I"~" " - AS4100 - 500 -..... 0 0 ©= 0.68 AISC - LRFD BS5950: Part 1 Eurocode 3 I I 20 40 I 60 ~ ~0 o Concentric loading • Eccentric loading I 80 100 L/r Fig. 7. C o r a p a r i s o n o f design s t r e n g t h s a n d test s t r e n g t h s for I-section c o l u m n s using n o m i n a l values. In the following comparison, the Eurocode3 adoption of the RondalMaquoi I~L approximations to the multiple ECCS 7 'a', 'b' and 'c' column curves is used as reference. Consequently, using a notation consistent with that of Section 3.1, the ECCS 'a', 'b' and 'c' curves are approximated K. J. R. Rasmussen, G. J. Hancock 48 closely by the slenderness reduction factor, 1 ~c ~+~~<1 (10) ~b=½ I-(1+ r/) + ~.2] (11) ~.--~f~ (12) n2E fE = (Lc/r)2 (13) r/= ~(2-0-2) (14) (0-21 'a' curve 0e = ~0-34 'b' curve {,0.49 'c' curve (15) where It follows from eqns (10)--(15) that for a given value of 0~, the slenderness reduction factor (~c(2)) is uniquely defined by the slenderness ().). The strength curves of the Australian, American, British and European specifications were obtained on the following basis: • • • In using the Australian Standard, the 0cb= - 0 . 5 curve was used in accordance with the recommendations made in Section 3.2 above. The ~b = --0"5 curve is a close fit TM to the 'a' curve of the ECCS Recommendations. The American Specification has not adopted the multiple column curve concept, but uses a single curve which is a fit to the SSRC 2P curve. 1° The SSRC 2P curve is based on a mean overall geometric imperfection of 1/1470 of the length, Is and lies between the 'a' and 'b' curves of the ECCS Recommendations at intermediate and long column lengths but below the 'b' curve at short lengths. The column curves of the British Standard are defined by eqns (10)--(15), except that the imperfection parameter (r/) is given by i/= O.O01a ~f~-~ (2-0.2) (16) Tests of high strenyth steel columns 49 rather than by eqn (14). The constant a takes the values 2.0, 3.5 and 5.5 for the British 'a', 'b' and 'c' curves respectively. Consequently, in using the British Standard, the slenderness reduction factor (0~c(2, fy)) is a function of the yield stress. The curve to be used for box and I-section (minor axis bending) columns fabricated from flame-cut plates (t < 40 ram) is the 'b' curve. For mild steel (fy = 235 MPa) this curve coJincides with the ECCS 'b' curve, but for high strength steel (fy = 690 MPa) the imperfection parameter (eqn (16)) becomes, r/=0.190(2-0.2) • (17) and so for this value of yield stress the British 'b' curve is nearly the same as the ECCS 'a' curve, as defined by eqns (14) and (15), alt]aough slightly higher. The Eurocode3 column curves are defined by eqns (10)--(15). The curve specified for welded box columns is the 'b' curve. For welded I-sections bent about their minor axis, the specified column curve is the 'c' curve. This curve is lower than those specified in AS4100 and BS5950, partly because I-sections fabricated from flame-cut plates may be designed using a higher column curve than sections fabricated from as-rolled plates according to AS4100 and BS5950, whereas no such distinction is made in Eurocode3. In summary, the column curves of the Australian, American and British specifications to be used in the comparison with test strengths all fit closely the ECCS 'a' curve. However, the column curves specified in Eurocode3 are the 'b' and 'c' curves for welded box sections and welded I-sections bent about their minor axis respectively. 4.2 Box c,olumns The design strengths are compared with test strengths in Fig. 6. Generally, the design strengths are in close agreement with the tests of the eccentrically loaded columns, although the Eurocode3 design curve is conservative at intermediate and long lengths. The design strength of the Australian Standard is slightly higher than the strength of test specimen Bll50E. However, the eccentricity (1.83 Lt/1000) used in this test was significantly higher them the nominal value of Lt/lO00. 4.3 I-section columns The design strengths are compared with the test strengths in Fig. 7. The Australian, American and British specifications generally agree with the 50 K. J. R. Rasmussen, G. J. Hancock tests, although the design specifications are conservative at short to intermediate lengths. The Eurocode3 design curve is significantly lower than the tests at intermediate and long lengths. This is partly because I-sections fabricated from flame-cut plates are designed using the same column curve as sections fabricated from as-rolled plates, as explained above. The test strengths of I1650E and I2950E are 22% and 18% higher than the design strengths respectively. 5 CONCLUSIONS A test programme on long box and I-section columns fabricated from high strength steel with a nominal yield stress of 690 MPa has been described. The purpose of the programme was to select a curve for high strength steel columns from the multiple column curves used in the Australian Standard. It was shown that the appropriate column curve for welded box columns with plate thickness less than 4 0 m m was the ~b = --0"5 curve. This was also the appropriate curve for welded I-section columns (minor axis buckling) fabricated from flame-cut plate with thickness less than 40 ram. The 0Cb= --0"5 curve of the Australian Standard fits closely the 'a' curve of the ECCS Recommendations. If the test specimens had been fabricated from ordinary steel, the column curve would have been the 0Cb= 0 curve according to the Australian Standard. This curve is lower than the ~b=--0"5 curve, and so the recommendation of using the ~b = --0"5 curve for high strength steel implies that columns fabricated from high strength steel are stronger than columns fabricated from ordinary steel when compared on a nondimensional basis. The design strengths of short box columns (and other cross-sections composed mainly of high strength plates supported along both longitudinal edges) may be optimistic if the slenderness of the component plates is near the yield slenderness limit. This is a result of the strength curves of the Australian, American and Eurocode3 specifications for high strength steel plates supported along both longitudinal edges which are optimistic for plate slenderness values near the yield slenderness limit. 6 This conclusion applies to a lesser extent to the British Standard which generally produces conservative strengths for plates supported along both longitudinal edges. The test strengths were compared with nominal design strengths using the Australian, 1 American, 2 British a and Eurocode35 specifications for steel structures. The design curves of the Australian, American and British specifications were nearly identical for welded high strength box columns and 1-sections bent about their minor axis, and were shown to be in close Tests of high strength steel columns 51 agreement with the test strengths. However, the Eurocode3 design curves were conservative c o m p a r e d with the tests. This was because the specification uses the ECCS 'b' curve for welded box sections, whereas curves similar to the higher ECCS 'a' curve are specified in the Australian, American and British specifications. The British curve closely fits the ECCS 'a' curve because the imperfection parameter is a function of the yield stress in the British Standard. F o r welded I-sections bent about their m i n o r axis, the Eurocode3 design curve was conservative because it was based on the ECCS 'c' curve whereas the Australian, American and British specifications were based on curves similar to the ECCS 'a' curve. In the comparison, the Eurocode3 design curves were obtained from Section 5.5.1 of Eurocode3. It is possible that changes to these rules m a y be specified in Annex D of Eurocode3. This annex applies to high strength steel members and is currently in preparation. ACKNOWLEDGEMENTS The authors wish to thank Bisalloy Industrial Steels Pty Limited for permitting release of the B I S A L L O Y 80 test results. REFERENCES 1. AS4100-1990, Steel structures. Standards Association of Australia, Sydney, 1990. 2. Load and resistance factor design specification for structural steel buildings. American Institute of Steel Construction, Chicago, 1993. 3. BS 5950: Part 1, Structural use of steelwork in building. British Standards Institution, London, 1990. 4. BS 436;0, Specification for weldable structural steels. British Standards Institution, London, 1990. 5. Euroeode No. 3: Design of steel structures, Part 1.1 General rules and rules for buildings. Draft Document ENV 1993-1-1: 1992, European Committee for Standardisation (CEN), Brussels, 1992. 6. Rasmussen, K. J. R. & Hancock, G. J., Plate slenderness limits for high strength steel sections, Journal of Constructional Steel Research, 23 (1992) 73-96. 7. European Recommendations for Steel Construction, Proc., European Convention for Constructional Steelwork (ECCS), Brussels, 1977. 8. Nishino, F. & Tall, L., Experimental investigation of the strength of T-1 steel columns. Fritz Engineering Laboratory Report No. 290.9. Lehigh University, 1970. 9. Tebedge, N. & Tall, L., Residual stresses in structural steel shapes. Fritz Engineering Laboratory Report No. 337.34. Lehigh University, 1973. 52 K. J. R. Rasmussen, G. J. Hancock 10. Galambos, T.V. (ed.), Guide to Stability Desiffn Criteria for Metal Structures, 4th edn. John Wiley and Sons, New York, 1988. 11. Abel, A. & Chung, Y. S., Effective stresses in cyclic deformation. Scripta Metallurffica, 13 (1979) 907-910. 12. AS1391-1974, Methods for tensile testing of metals. Standards Association of Australia, Sydney, 1974. 13. Rondal, J. & Maquoi, R., Single equation for SSRC column-strength curves. Journal of the Structural Division, ASCE, 105 (ST1), (1979) 247-50. 14. Rotter, J.M., A simple approach to multiple column curves. Proc., Metal Structures Conference, Newcastle, Australia, 1981. 15. Bjorhovde, R., Deterministic and probabilistic approaches to the strength of steel columns. PhD Thesis, Lehigh University, Bethelehem, PA, 1972.