Experimental investigation and design of aluminum columns with longitudinal welds

Abstract

This paper presents an experimental investigation of longitudinally welded aluminum alloy I-section columns subjected to pure axial compression. The specimens were fabricated using 6061-T6 heat-treated aluminum alloy. The test program included 20 column tests which were separated into 2 test series of different types of welding sections. Each test series contained 10 columns. All the specimens were welded using the Tungsten Inert Gas welding method. The length of the specimens ranged from 442 to 2433 mm in order to obtain a column curve for each test series. The observed failure mode for the column tests includes mainly flexural buckling around the minor axis and the major axis by applying support except for one column (ZP1217-1) which buckled in the local zone and some columns which failed in the weld. The test strengths were compared with the design strengths predicted by the European Code and China Code for aluminum structures. The purpose of this paper is to present the tests results of two typically longitudinally welded I-section columns, and to check the accuracy of the design rules in the current specifications.

Keywords

aluminum alloy / longitudinal weld / heat-affected zone (HAZ) / reduced strength zone (RSZ) / buckling / column / experimental study

Introduction

Aluminum alloy as a new type of structural material has the characteristics of lightness, corrosion resistance and beautiful shape. Thus, it is becoming more and more widely used in modern structures. Aluminum members are normally manufactured by heat-treated aluminum alloys because heat-treated aluminum alloys have notably higher yield stress than non-heat-treated alloys. However, when heat-treated aluminum alloys are welded, the heat generated from the welding reduces the material strength significantly in a localized region, and this is known as heat-affected zone (HAZ) softening. The effects of welding on the strength and behavior of aluminum structural members depend on the direction, location and number of welds. In aluminum structures welds are divided into two types, namely (1) transverse welds and (2) longitudinal welds, for the purpose of dividing their influence on member strength. The current American Aluminum Design Manual [1], British Code BS8118 [2] and European Code [3] for aluminum structures provide design rules for structural members containing transverse and longitudinal welds. In China, the Aluminum Design Code is newly published. Though the behavior of non-welded columns has been extensively investigated [4-7], few studies have been performed on the welded columns. Thus, the purpose of this paper is first to present a test program on two typically longitudinally welded I-section columns, and second, to compare the test strengths with the design strengths predicted using the European Code and China Code for aluminum structures.

Experimental conditions

The description of the material specimens

The material in the test is weak hardening aluminum 6061-T6 which is often used in structures. All the welded tensile coupon specimens were welded by Tungsten Inert Gas (TIG) welding method. The filler metal is 5A06 aluminum alloy. The depth of the plate is 6 mm. This welding procedure is identical to that of the column specimens. To determine the width of the HAZ and the material properties in the different parts of the section explicitly, two batches of tests were performed. In the first batch of tests, tensile [8] and Vickers hardness tests [9] were performed to determine the width of HAZ. The specimens were perpendicular to the weld axis as shown in Fig. 1, in which the weld is between the two black lines. On the basis of the results of the first batch, the second batch of tests, including three kinds of material specimens which were made in the parent, weld zone and the HAZ respectively as shown in Fig. 2 [10], was performed to determine the material properties explicitly.

The test results of the material specimens

Through the first batch of test, the half width of the HAZ, i.e., b_{\mathrm{H}\mathrm{A}\mathrm{Z}}=\mathrm{20}\mathrm{m}\mathrm{m} is determined, which is the scope of the strength near the weld changing from the lowest to the same as the parent material, as shown in Fig. 3. At the same time, the half width of the reduced strength zone (RSZ), i.e., b_{\mathrm{r}}=\mathrm{14}\mathrm{m}\mathrm{m} is determined by calculating. The RSZ is the zone where strength reduces strikingly when the reduction of strength in curve distributing in the HAZ is equal to that in rectangle distributing. The corresponding stress is called the reduced elastic limit stress, usually taking the value of the proof yield stress f_{\mathrm{0.2}}^{*} of the heat affected material. The corresponding calculating format is

f_{\mathrm{0.2}}^{*}{b}_{\mathrm{r}}+f_{\mathrm{0.2}}(b_{\mathrm{H}\mathrm{A}\mathrm{Z}}-b_{\mathrm{r}})={\displaystyle {\int }_{\mathrm{0}}^{b_{\mathrm{H}\mathrm{A}\mathrm{Z}}}f(x)dx}.

Thus, the half width of RSZ is determined as

b_{\mathrm{r}}={\displaystyle \frac{f_{\mathrm{0.2}}{b}_{\mathrm{H}\mathrm{A}\mathrm{Z}}-{\displaystyle {\int }_{\mathrm{0}}^{b_{\mathrm{H}\mathrm{A}\mathrm{Z}}}f(x)dx}}{f_{\mathrm{0.2}}-f_{\mathrm{0.2}}^{*}}}.

For the second batch of specimens, due to their small dimensions, the deformation was recorded by a customized electronic extensometer with 15 mm gage length, and the load was applied at a constant speed of 1 mm/min [11]. With the help of the computer acquisition system, the stress-strain curve is obtained, as shown in Fig. 4. By using Origin software, the parameter n in the Ramberg-Osgood expression [12] used to denote the material relation is fitted. The test results of material properties are listed in Table 1. AL, WELD and HAZ indicate the parent metal, welded metal and HAZ metal, respectively.

Brungraber and Clark material tests [13]

In Brungraber and Clark test, the longitudinally welded column specimens were made from 1/2in. thick plate of alloys 6061-T6, 5154-H34, and 5456-H321. The average mechanical properties are listed in Table 2. The test results of the hardness surveys are listed in Table 4.

The description of the tests

The tests in the paper

In the paper, two typically welded I-sections with the same nominal section size 102 mm × 66 mm × 6 mm × 6 mm columns were studied. One is called the T-shape welded specimen and the other is called the P-shape welded specimen. The T-shape welded specimens were welded from flat strips. The P-shape welded specimens were welded from double T extruded sections and one flat strip, as shown in Fig. 5. The flat strips and the T extruded profiles were supplied by the manufacturer in uncut lengths of 6000 mm. The flat strip and T extruded profile were cut to a specified length ranging from 282 to 2273 mm. The dimension and the number of the specimens are listed in Table 3. The calculating length L₀ is the actual length L plus the thickness of a pair of pinned bearings, as shown in Fig. 6, i.e., L₀ = L + 160 mm. The dimension and the number of the column specimens in Brungraber and Clark are listed in Table 4.

The purpose of the tests in the paper is first to study the ultimate capacity of the longitudinally welded aluminum columns; and second, to compare the test strengths with the design strengths predicted using European Code (EC9) and the Code for Design of Aluminum Structures of China after modification of A_{\mathrm{H}\mathrm{A}\mathrm{Z}} to A_{\mathrm{R}\mathrm{S}\mathrm{Z}} respectively, and finally, to validate the applicability of the formula for the longitudinally welded column in the China Code after modification. The load was applied through the reaction frame and hydraulic pressure jack. The bilateral edge hinge support corresponding to pinned-end bearing, as shown in Fig. 6, was introduced at the end of each column. The pinned end columns were centered with the aid of six strain gauges which were symmetrically attached around the section at the mid-length of the column. The same gauges were used for measuring the bending strains. Four displacement transducers were placed at the mid-length of the specimen to measure the lateral deflection. The arrangement of the strain gauges and displacement transducers can be seen in Fig. 7. Eight displacement transducers were placed at the ends of each column to measure the axial shortening.

The method of loading: the loads were applied step by step. The load in each step was 0.2P_u (P_u is the ultimate load). When the load was near the ultimate load, the load in each step was changed to 0.1 P_u. Little change was made to the specimen to let the strain gauge reading in the first load step near equivalency. The column was considered to be centered until each strain gauge reading was no more than 5% of their average. When each strain gauge reading became stable, a data acquisition system was used to record the applied load and the readings of the displacement transducers, as well as the strain gauge readings at regular intervals during the tests. A continuous record of load versus strain was obtained. A typical one is shown in Fig. 8.

The tests of Brungraber and Clark

The column specimens were welded by plates with two ends fixed. The detailed information of the specimens and the test results are listed in Table 4.

The relative design specifications

The specification in the European Code for the ultimate capacity of longitudinally welded columns [3]

In the code, the ultimate load N_{b,Rd}, of a longitudinally welded column is given by modifying the stability coefficient of the unwelded column, that is, through multiplying the stability coefficient \chi of the unwelded column by modified coefficient k_{\mathrm{2}}. Then N_{b,Rd}, may be determined from

(1)N_{b,Rd}=f_{s}A/r_{M1}=\eta k_1{k}_2\chi f_{0.2}A/r_{M1},

where \chi is taken from Fig. 6.8 (Ref. [3])

(2)

\varphi =\mathrm{0.5}\left(\mathrm{1}+\alpha \left(\overline{\lambda }-{\overline{\lambda }}_{\mathrm{0}}\right)+{\overline{\lambda }}^{\mathrm{2}}\right), \alpha is an imperfection factor.

For weak hardening aluminum,

(3)\alpha =0.20,{\overline{\lambda }}_0=0.10\overline{\lambda }={\displaystyle \frac{\lambda }{\mathrm{\pi }}}\sqrt{{\displaystyle \frac{\eta f_{0.2}}{E}}},

\eta is the factor to allow for local buckling; k_{\mathrm{1}} is the factor to allow for the asymmetry of the cross-section, for symmetric cross-section, k_{\mathrm{1}}=\mathrm{1}; f_{\mathrm{0.2}} is the nominal yield strength for the parent metal; A is the gross area; k_{\mathrm{2}} is a factor to allow for the weakening effects of longitudinal welding. For weak hardening aluminum,

(4)k_{\mathrm{2}}=\mathrm{1}-\left(\mathrm{1}-{\displaystyle \frac{A_{\mathrm{1}}}{A}}\right){\mathrm{10}}^{-\overline{\lambda }}-\left(\mathrm{0.05}+\mathrm{0.1}{\displaystyle \frac{A_{\mathrm{1}}}{A}}\right){\overline{\lambda }}^{\mathrm{1.3}\left(\mathrm{1}-\overline{\lambda }\right)}\mathrm{.}

In which, A_{\mathrm{1}}=A-A_{\mathrm{H}\mathrm{A}\mathrm{Z}}(\mathrm{1}-{\rho }_{\mathrm{H}\mathrm{A}\mathrm{Z}}), A_HAZ = area of HAZ, A_{\mathrm{H}\mathrm{A}\mathrm{Z}}={\displaystyle \sum _i{b}_{\mathrm{H}\mathrm{A}\mathrm{Z},i}}{t}_i; b_{\mathrm{H}\mathrm{A}\mathrm{Z},i} is the half width of HAZ, t_i is the depth of the plate.

{\rho }_{\mathrm{H}\mathrm{A}\mathrm{Z}} is the reduction factor of the yield strength in the heat affected material. When the design is dominated by the 0.2% proof strength, {\rho }_{\mathrm{H}\mathrm{A}\mathrm{Z}}={\displaystyle \frac{f_{\mathrm{0.2}}^{*}}{f_{\mathrm{0.2}}}}.

The design specification in China Code for the ultimate capacity of longitudinally welded aluminum columns [14]

The following formula is adopted in the Code for Design of Aluminum Structures in China,

(5){\displaystyle \frac{N}{\overline{\phi }A}}\le f,

(6)\overline{\phi }={\eta }_{\mathrm{e}}{\eta }_{\mathrm{H}\mathrm{A}\mathrm{Z}}\phi ,

{\eta }_{\mathrm{e}} is the factor to allow for local buckling, {\eta }_{\mathrm{e}} is equal to \eta in the European Code, A is hte gross area.

(7)\phi =\left({\displaystyle \frac{1}{2{\overline{\lambda }}^2}}\right)\left\{\left(1+\eta +{\overline{\lambda }}^2\right)-{\left[{\left(1+\eta +{\overline{\lambda }}^2\right)}^2-4{\overline{\lambda }}^2\right]}^{1/2}\right\},\mathrm{a}\mathrm{n}\mathrm{d}\phi \le 1,

(8)\eta =\alpha \left(\overline{\lambda }-{\overline{\lambda }}_{\mathrm{0}}\right),

for weak hardening aluminum member, \alpha =\mathrm{0.2}, {\overline{\lambda }}_{\mathrm{0}}=\mathrm{0.15};\overline{\lambda }={\displaystyle \frac{\lambda }{\pi }}\sqrt{{\displaystyle \frac{{\eta }_{\mathrm{e}}{f}_{\mathrm{0.2}}}{E}}}.

{\eta }_{\mathrm{H}\mathrm{A}\mathrm{Z}} is the reduction factor allowing for the residual stress and the welding heat effect. The formula is basically the same as k_{\mathrm{2}} in the European Code, in which, \overline{\lambda } and A_{\mathrm{1}} are used to consider the effects of the residual stress of welding and welding softening respectively. On the basis of referring to the related literatures [13,15] and comparing with the test results, it is suggested that A_{\mathrm{H}\mathrm{A}\mathrm{Z}} is changed to A_{\mathrm{R}\mathrm{S}\mathrm{Z}} in the paper. A_{\mathrm{R}\mathrm{S}\mathrm{Z}}={\displaystyle \sum _i{b}_{\mathrm{r},i}{t}_i}, b_{\mathrm{r},i} is the half width of the RSZ.

If the effect of the residual stress of welding is negligible, \overline{\lambda }=\mathrm{0}, then

(9){\eta }_{\mathrm{H}\mathrm{A}\mathrm{Z}}={\displaystyle \frac{A_{\mathrm{1}}}{A}}=\mathrm{1}-(\mathrm{1}-{\rho }_{\mathrm{H}\mathrm{A}\mathrm{Z}})\times {\displaystyle \frac{A_{\mathrm{R}\mathrm{S}\mathrm{Z}}}{A}}.

The test results

Without the residual stress

In the tests of Brungraber and Clark, each specimen consisted of two plates joined by a butt weld to form a solid rectangular section. After being welded, the plate specimens were sawed to size. During sawing the column specimens out of the larger welded plates, the residual stresses were considerably relieved. It was measured that the residual stresses were very small and their effects could be neglected [13]. The values of {\eta }_{\mathrm{H}\mathrm{A}\mathrm{Z}} obtained from the test results and by calculating are listed in Table 4 column 10 and column 11. The values of load ratio are listed in Table 4 column 12. It can be seem that they are in good agreement with each other.

With the dual stress

In the paper, it was verified by the tests that the residual stress was very large and could not be neglected [16]. Furthermore, no measure such as annealing was taken to relieve the stress in the members. The main failure mode of the specimens was flexural buckling around the minor axis and the major axis by applying support (Fig. 9) except for the column ZP1217-1 which buckled in the local zone (Fig. 10) and some columns which failed in the weld (Fig. 11).

The ultimate capacity test value of each specimen is listed in Table 5 column 4 and 10. The values of load ratio P_{\exp }/P_{\mathrm{C}\mathrm{H}\mathrm{I}\mathrm{N}\mathrm{A}} and P_{\exp }/P_{\mathrm{E}\mathrm{C}\mathrm{9}} are listed in Table 5 column 5, 6 and 11, 12. The comparison of the test value with the design value of the Codes is also shown in Fig. 12. From Table 5 and Fig. 12, it can be seen that the design value given by the European Code and China Code when A_{\mathrm{H}\mathrm{A}\mathrm{Z}} was changed to A_{\mathrm{R}\mathrm{S}\mathrm{Z}}, is in good agreement with that of the test for the T-shape welded specimens. While for the P-shape welded specimens, the effect of welding to the capacity of the specimen was small because the weld was near the neutral axis of the section and the design strengths predicted by the two Codes are conservative. Generally, the design strength predicted by the European Code is more conservative than by the China Code when the slenderness ratio is less than 1. For the P-shape welded specimens, the mean values of the load ratio P_{\exp }/P_{\mathrm{C}\mathrm{H}\mathrm{I}\mathrm{N}\mathrm{A}} and P_{\exp }/P_{\mathrm{E}\mathrm{C}\mathrm{9}} are 1.25 and 1.27, with the corresponding COV (coefficient of variation) of 0.161 and 0.165, respectively. For the T-shape welded specimens, the mean values of the load ratio P_{\exp }/P_{\mathrm{C}\mathrm{H}\mathrm{I}\mathrm{N}\mathrm{A}} and P_{\exp }/P_{\mathrm{E}\mathrm{C}\mathrm{9}} are 1.08 and 1.09, with the corresponding COV of 0.086 and 0.085, respectively.

Conclusions

The tests of hardness and tensility were performed, thus the half width of HAZ and RSZ of the test specimens was determined. The material properties of the different parts of the welded specimens were explicitly determined. A test program on two typical types of longitudinally welded I-section pinned-end aluminum columns has been presented. A comparison of the column test strengths with the design strengths predicted by the European Code and China Code after modification has been presented. It is verified that the test results were in good agreement with that of the design by changing A_{\mathrm{H}\mathrm{A}\mathrm{Z}} to A_{\mathrm{R}\mathrm{S}\mathrm{Z}} in the two codes for the solid rectangular section specimens joined by a butt weld and for the T-shape section specimens with fillet weld. Whereas for the P-shape section specimens with groove butt welds, the design strengths predicted by the two Codes are conservative. Generally, the strength predicted by the European Code is more conservative than by the China Code when the slenderness ratio is less than 1.

References

[1] A A. Aluminum design manual. Washington (DC): The Aluminum Association, 2005

[2] BS 8118: Part 1: British Standards: Structural use of aluminium-Part 1: Code of practice for design, BSI Standards, London, 1991

[3] BS EN 1999–1-1:2007, Eurocode 9: Design of aluminium structures-Part 1–1: General structural rules, European Committee for Standardization, Brussels, 2007

[4] Xin J. Study on Resistance of Uniaxial Symmetrical Aluminum Alloy Compression Members. Dissertation for the Doctoral Degree. Shanghai: Tongji University, 2005 (in Chinese)

[5] Wu Y G. Study on Local Buckling of Aluminium Elements And Flexural-Torsional Buckling of Aluminium Beams. Dissertation for the Doctoral Degree. Shanghai: Tongji University, 2006 (in Chinese)

[6] Zhang Z. Study on Stability Capacity of Aluminium Beam-Columns. Dissertation for the Doctoral Degree. Shanghai: Tongji University, 2006 (in Chinese)

[7] Guo X N. Theoretical and Experimental Research on Aluminum Alloy Structure Members. Dissertation for the Doctoral Degree. Shanghai: Tongji University, 2006 (in Chinese)

[8] GB2651–89. Method of tensile test for welded joint, 1989 (in Chinese)

[9] GB/T 4340–1999. Metallic vickers hardness test, 1999 (in Chinese)

[10] GB 2649–89. Methods of sampling for mechanical properties tests of welded joint, 1989 (in Chinese)

[11] GB2652–89. Methods of tensile tests for weld and deposited metal, 1989 (in Chinese)

[12] Ramberg W, Osgood W R. Description of stress-strain curves by three parameters, NACA Techn No902, 1943

[13] Brungraber R J, Clark J W. Strength of welded aluminium columns. Transaction of ASCE, 1962, 127(2): 202-225

[14] GB50429–2007. Code for design of aluminium structures, 2007 (in Chinese)

[15] Mazzolani F M. Aluminium alloy structures (2nd edition). London: E&Fn Spon, 1995

[16] Wu Y, Zhang Q L. Experimental study on residual stresses of welded aluminium members.Steel Structure, 2007, 22(92): 29-32

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg