Moment resistance of of strong-axis open-section beam-to-column welded steel joints

This Report has been developed in the framework of a collaboration between ISISE (Institute for Sustainability and Innovation in Structural Engineering) at The University of Coimbra  and IDEA StatiCa. This project has been funded jointly by both entities and authored by:

• Luis Simões da Silva
• Jorge Conde
• Filip Ljubinković
• João Pedro Martins
• Francisca Santos
• Fernando Freire
• Juan Aguiar

Executive summary

This report presents a thorough analysis of welded beam-to-column steel joints with a specific focus on the column web in shear component, using properly validated, high-quality 3D-FEM models. The study comprises strong-axis moment-resisting welded open-section beam-to-column joints, with and without transverse stiffeners, covering one-sided and two-sided joints located at either an intermediate (internal) or last story (roof), and, for internal joints, with different levels of column axial load. The study comprises a wide range of situations in terms of column slenderness, joint aspect ratio, joint configuration, stiffeners, axial force, and moment ratio (for two-sided joints). FE-relevant aspects, such as material model, analysis type, mesh density, initial imperfections, etc., are also considered.

This study comprises a year of work in the following areas:

1. The set list includes cases with 30%, 50%, and 70% axial load.
2. The steel grades include S235, S275, and S355.
3. Rolled columns, fillet- and butt-welded columns are discussed.
4. A comprehensive study on the differences between GMNA and MNA has been included.
5. A comprehensive study on the influence of the material constitutive law has been included.
6. The influence of fillet welds on the joint behavior is studied.

Objectives

The objective of this study is the assessment of strong-axis bending-moment resistance of welded joints in one- and two-sided beam-to-column configurations with open-section members (beams and columns), including the influence of transverse stiffeners, using: i) European design rules, ii) FEA (3D solid elements) and iii) IDEA StatiCa; and the comparison of results obtained by each method.

Methodology

1. Assess the design resistance of the web panel component for a wide range of welded beam-to-column joints through high-quality 3D FEM models, used subsequently as benchmark cases.
2. Compare the joint resistance with CBFE models (IDEA StatiCa), based on shell elements.
3. Discuss the possibility of improving the features and default options offered by IDEA StatiCa to ensure that a realistic and safe value of resistance is obtained.
4. Compare the joint resistance with results obtained from design expressions provided in Eurocode 3, current and forthcoming versions.
5. Assess the differences in moment resistance produced by the 5% equivalent plastic strain criterion adopted by IDEA StatiCa to define the ‘design resistance’ on the moment-rotation curve, versus other common criteria (such as the relaxation of the secant stiffness to 1/3 of the initial stiffness).

Questions addressed

• What is the scatter of results?
• What is the influence of the different parameters (slenderness, aspect ratio, bending moment ratio, transverse stiffeners, axial force, etc.) on these results?
• What is the influence of the column web components on these results?

Exclusions

No reliability assessment of program results is carried out as it is outside the scope of this study.

Summary of conclusions

The analysis of the Abaqus results has highlighted the following conclusions:

1. Regarding the 5% equivalent plastic strain criterion, it has been shown that, for this type of joint, it produces resistance results like those obtained from the same numerical model with a reduction of secant stiffness to 1/3 of the initial stiffness.
2. Initial imperfections play a minor role in the behavior of the joints analyzed, however, geometrical non-linearity (2nd order analysis) should be included, particularly in cases where the axial load is present.
3. Inclusion of strain hardening results in a small average increase of 4% in resistance, however, it has a large effect, about 44% on average, on the joint rotation.

The following comments are relevant for implementation in IDEA StatiCa:

1. IDEA StatiCa, with the program’s default mesh (version 23.1) and MNA (materially non-linear analysis), provides resistances that are larger than those obtained with Abaqus. On average, 6% larger for rolled columns with butt welds, and about 11% for welded columns with butt or fillet welds. For rolled columns with fillet welds, the results are, on average, identical to those of Abaqus. The peak deviations observed for an individual case are 22% (rolled columns with butt welds) and 53% (welded columns with butt welds).
   IDEA StatiCa comment: This situation is remedied by two changes implemented in version 25.0:
   - Geometrically nonlinear analysis
   - Weld spreading area – reduced for butt welds, i.e., resistance of butt-welded columns is decreased
   
2. IDEA StatiCa with a refined mesh (halving the size of the default mesh) and GMNA (geometrically and materially non-linear analysis) provides resistances that are in good accordance (within ±5% on average) to the Abaqus results. The peak deviations observed for an individual case are 15% (rolled columns with butt welds), 5% (rolled columns with fillet welds), 27% (welded columns with butt welds) and 14% (welded columns with fillet welds).
   IDEA StatiCa comment: Halving the mesh is no longer necessary with the changes mentioned in the previous point.
   
3. The influence of initial imperfections was studied for one-sided internal joints and found to be negligible.
   IDEA StatiCa comment: This means GMNA is sufficient and GMNIA is not necessary for the investigated range of joints.
   
4. Purely material non-linear analysis without 2^nd order effects leads to unconservative results in case of axial force on the column. Therefore, it is recommended that the geometrically nonlinear analysis is activated by default in the program.
   IDEA StatiCa comment: In version 25.0, Geometrically nonlinear analysis is made available for every connection. In the upcoming patch, a warning will be provided.
   
5. The mesh size plays an essential role in the accuracy of results. A potential approach to estimate the converged value could be based on an automated process that halves the mesh size and applies the Richardson extrapolation to the moment resistance obtained with both values.
   IDEA StatiCa comment: Richardson extrapolation is possible when determining the load resistances. However, it is not possible for passed/failed checks. With the previously-mentioned changes in version 25.0, the default mesh already provides sufficiently accurate results. 
6. The material model implemented by IDEA StatiCa (with E/1000 strain hardening slope) results in an average of 3% increase in moment resistance when compared to the results obtained with elastic perfectly plastic material.

Read the full report in the attachment.

Anhänge zum Download

• 01 Report IDEA STATICA FINAL SIGNED.pdf (PDF, 10,7 MB)