CHAPTER 2 PREVIOUS RESEARCH AND DESIGN FORMULATION
2.1 R EVIEW OF RESEARCH ON AS - WELDED CHS JOINTS
Extensive research has been conducted both experimentally and numerically on as- welded CHS joints over the last five decades. As a result, the design formulations are progressively updated to incorporate the larger geometric ranges and the newly discovered failure modes which are the main concerns of most researchers.
2.1.1 Experimental research
Experimental research on the strength of tubular joints was initiated in the early 1950s in the University of Texas and University of California. Only a few tests were carried out in
this period (Toprac, 1961; Washio et al, 1968). The joint configurations and the geometric ranges were also quite limited in these tests. Based on the limited results from his tests, Toprac (1966) first investigated the effects of the parameters α, β and γ on the strength of an as-welded joint and observed some tremendous reserve strengths in simple tubular joints.
In 1970s, there was a rapid development of the research on tubular joints. Many tests were conducted and significant results under a wide range of geometric parameters were released. The efforts at that time were focused on simplified techniques in obtaining elastic stress distributions. Pan et al (1976), summarized the failure patterns of simple uni-planar tubular X-, T- and K- joints on the basis of the test results reported by the previous researchers and concluded that there were seven possible failure modes for X- and T- joints, namely, brace tensile failure, weld tensile failure, tensile crack in the chord, plastic deformation of the chord, chord wall buckling, lamellar tearing of thick wall joints and local collapse of chord wall. As for K-joints, two additional failure modes were observed, which were the crack failure at the weld toe of the tensile brace and the brittle tensile failure of the chord.
Later, Yura et al (1980) summarized the results of 137 ultimate strength tests on simple uni-planar tubular joints and proposed a set of ultimate capacity equations for different types of joints under brace axial, IPB and OPB loading, together with the suggestion of a deformation limit for different loading cases to determine the joint strength. Table 2-1 shows the geometry range in Yura’s database.
Yura’s capacity equations have served as the basis for the CIDECT and IIW design guidance. However, due to the limitation of the test rig, most of the tests in 1970s were
conducted using quite small specimens, which were the main resources for Yura to draw his conclusions. Thus, with more results from large scale tests available in 1980s, an even larger joint database (747 joint tests, as shown in Table 2-2) was reviewed by Kurobane et al (1984) and by introducing two screening criteria to guarantee the reliability of the resources, a new set of design equations for X-, T- and K joints were developed by him.
Table 2-1 Geometry range for Yura’s database
Configuration Load type D0 (mm) β γ θ X- Axial 140-457 0.19-1.00 10.3-47.6 90o
Axial 140-456 0.17-0.84 10.8-46.5 90o IPB 219-457 0.19-0.81 11.0-47.6 90o T-
OPB 165-507 0.19-0.90 18.4-47.6 90o Axial 165-456 0.17-0.84 17.5-46.5 90o
Y- OPB 507 0.9 20.8 90o
Axial 165-508 0.17-0.69 13.8-51.6 30o-90o
IPB 507 0.64-0.90 22.2 30o-90o
K-
OPB 507 0.64-0.90 22.8 30o-90o
Table 2-2 Geometry range for Kurobane’s database (brace axial load)
Configuration Load type D0 (mm) β γ θ X- Axial 60-1400 0.19-1.00 6.5-49.0 60o-90o T-/Y- Axial 60-1400 0.19-1.00 8.5-49.5 45o-90o K- IPB 60-1400 0.19-1.00 7.5-51.0 30o-90o
Kurobane et al. (1986) reported the study on the local buckling behavior of CHS K-joints under brace axial loading. Eight overlapped K-joints and three gap K-joints were loaded up to failure. The local buckling strength formulation was derived from these test results, which recommended that the brace diameter to thickness ratio should be limited to 0.1E/Fy.
After the basic aspects tubular joints were investigated extensively and also with the improvements in test techniques, the focus of research was turned to the behavior of the tubular joints under practical conditions, like combined loadings or failure patterns.
Stamenkovic and Sparrow (1983) reported the test results on the load interaction behavior of CHS T-joints, which is the first study on the brace load interaction for the CHS T- joints. A linear relationship between the brace axial load and IPB for the joints with different β ratios was revealed. Makino et al. (1986) also conducted a series of tests on 25 T-joints and 10 K-joints under combined brace loads. The interaction between the brace axial load and the OPB moment for T-joint can be represented by a straight line, the compressive chord stress induced for K- joints under brace IPB can be included in the chord stress function proposed by Kurobane (1984).
Multi-planar tubular joints were hardly investigated in the early research due to technical limitations. Paul et al. (1993) started to study the behavior of this type of joints by carrying out tests for 20 multi-planar TT and V-joints and a new equation for the strength of the TT joints was proposed. Makino and Kurobane (1994) presented the results of 9 KK-joint tests and concluded that the ultimate KK-joint capacity was governed by the local deformation of the chord wall.
The effect of the chord stress was first investigated by Togo (1967) and it was found that the effect of tensile chord stress was minor. However, the specimens adopted by Togo’s investigations were very small (D0=101.6mm). Later, ten large scale X-joint tests under the chord axial and IPB stress were reported by Boone et al. (1982), with three brace loading conditions (axial, IPB and OPB) and one β ratio (0.67). Weinstein and Yura (1985) extended Boone’s investigations with larger geometry range (β=0.35 and 1.0).
The effect of compression chord stresses on X-joints subjected to brace axial compression was studied by Kang et al. (1998). Thee X-joints with β=0.52 and γ=11.6 were tested in their investigations.
2.1.2 Numerical research
As 2.1.1 shows, a large amount of tests have been carried out regarding most major aspects of tubular joints. However, it is impractical to conduct tests to cover all types and configurations of tubular joints due to the economic and geometric restrictions. Finite Element (FE) method provides an excellent alternative for the experimental investigation.
The idea of FE was originally present in early 1960s. But due to the restriction from the computing capacity, the numerical (FE) research was not initiated until late 1970s, when there was a breakthrough in computer industry.
Hoffman et al (1980) first applied the FE method to the research of tubular joints. They investigated the effects and feasibility of different finite elements types and concluded that good predictions could be achieved if the technique was properly used.
Ever since that, FE method has been applied to every aspect of the research on tubular joints extensively. In 1995, van de Valk presented an extensive numerical investigation on the uni-planar and multi-planar X- and T-joints. He modeled the weld with one additional layer of shell elements along brace-chord intersection and the comparisons with the test results showed that this simplification could achieve sufficient accuracy.
Dexter and Lee (1999) carried out a systematic study on the behavior of axially loaded K- joints through FE method. They suggested that the chord length parameter α should be taken as 14 to avoid the chord ends effect. Besides, they proposed a set of criteria for the determination of the ultimate strength of a tubular joint in the FE analyses. These criteria
included: first peak load in load-deformation curves, brace squash load, a maximum plastic strain limit in element (0.2) and Yura’s deformation limit. These criteria proved to have the ability to define the joint strength with sufficient accuracy.
Cofer and Jubran (1992) present a new numerical approach in analyzing tubular joints subjected to tensile brace load. To include the possible fracture behavior of tubular joints, they incorporated a continuum damage mechanics approach into normal FE analyses. A damage variable was introduced to evaluate the failure of materials. The key damage parameters were determined by reproducing the results of a standard tensile coupon test.
Their approach was verified by a series of X- and T-joints and very close correlation between test and FE results was observed.
All these prove the effectiveness of numerical investigations on tubular joints. However, the limitation of FE method, like the failure of predicting the crack initiation and hence giving an over-estimated ultimate strength, need to be noted.