Evaluation and Optimisation of Adhesive Point Fixings in Structural Glass

01 January 2013 → 31 December 2016
Regional and community funding: IWT/VLAIO
Research disciplines
  • Engineering and technology
    • Structural design
    • Structural optimisation
    • Building technology
    • Construction materials
    • Construction materials technology
    • Construction mechanics
    • Structural engineering
    • Polymers and plastics
glass adhesives glass structures adhesively bonded point fixings
Project description

Its main property, i.e. transparency, has ensured that structural glass is used extensively throughout the world and has become an integral part of the skyline of major cities and constructions in general. Nonetheless, connections in structural glass members still represent a field of research of structural glass engineering which is developing and expanding rapidly. Traditional systems to connect glass to the supporting substructure and to improve the overall transparency, often consist of so-called point-fixings. With the current state of technology, it is common to use bolted solutions for point-fixings in glass. However, adhesive point-fixings have several advantages compared to bolted point-fixings, such as the stress redistributing ability of the adhesive, no weakening or residual stresses in the glass due to glass perforation, prevention of a thermal bridge, etc. These advantages, together with promising strength values are important reasons why adhesive bonding is currently considered a very promising alternative for structural glass point-connections. In this work a design method for glass panels supported by adhesive point-fixings is proposed by means of a flowchart. The design is based on the time-efficient SLG-method (Superposition of Local and Global components), developed by Beyer for the design of bolted point-fixings. In the flowchart, the design is divided in two parts, i.e. the glass design and the adhesive design. In the former, the glass panel is separated in a global component and a local component. Due to the separation into one global component that can be built up with a less dense mesh pattern and one local component that is built up with a more complex and dense mesh pattern, the stress distribution can be determined in a very time-efficient manner. The stress and deformation in the field are examined by means of the global model. The stress in the vicinity of the connection is examined by the sum of the global stress and the maximum principal stress in the local models. For the adhesive design, only one local model can be considered, i.e. the multi-axial model. By applying the obtained failure criteria the adhesive layer can be examined for the considered loads. The occurring stress in the glass panel and in the adhesive layer can be altered by chancing geometrical parameters and material parameters. When both designs are satisfied, the design of glass panels supported by adhesive pointfixings for the considered configuration is completed. The suitability of the SLG-method for adhesive point-fixings is examined by a FEA comparison between the stress distributions conducted on the one hand by the SLG-method and on the other hand by a FEA model in which the total glass plate with the adhesive connections is built up in detail with volumetric elements. Despite small deviations, the SLG-method predicts the occurring stresses in a glass panel supported by adhesive point-fixings conservatively and accurately. The stress distribution in the adhesive layer can also be determined by the SLG-method. As the stress distribution consists of only the sum of the local stresses in the SLGmethod, these stresses are derived from the multi-axial model. The local model is numerically validated for different geometrical properties (three glass thicknesses and three diameter connectors), different material properties (two adhesive types) and three different load conditions (tension, shear and multiaxial load). By supporting the glass panel along a circumference with a diameter equal to six times the connector diameter, the deformation of the glass panel is also taken into account. This deformation causes important stress concentrations in the adhesive layer. The multi-axial local model can be used to directly determine the sum of the local stresses. The material models in the local model are obtained for a rubber-like adhesive (MS-polymer Soudaseal 270 HS) and a glassy adhesive (two-component epoxy 3M™ Scotch-Weld™ 9323). Two-component epoxies and acrylates are thermosetting adhesives. The behaviour of these materials is typically elastic until failure and will fail at relatively small strains by the initiation and propagation of a crack. However, many adhesives are rubber-like materials, such as silicones and MS-polymers. Local deformations of the small-scale test specimens were measured through 3D-DIC. The comparison between the data from the test machine and the DIC-output revealed major differences between the measured deformations. This confirms that the use of DIC is needed to accurately measure the occurring strains during the small-scale tests. The total model for the validation of the SLG-method is experimentally validated. The experimental results demonstrate that the highest stresses are reached with the smallest edge distances. As expected the deformations are significantly larger with small edge distances. The numerical analyses show stress peaks which are not visible in the experiments. This highlights the benefits and necessity of numerical investigation. To reduce the stress in the glass panel or adhesive layer following actions can be taken, in order of decreasing influence:

- Increase the number of connections;

- Decrease the panel size;

- Increase the edge distance;

- Decrease the eccentricity;

- Increase the connector diameter;

- Increase the glass thickness;

- Decrease the Young’s modulus of the adhesive;

- Change the boundary condition to hinged;

- Increase the adhesive thickness;

- Decrease the Poisson ratio of the adhesive.

Furthermore, the failure criteria for the two selected adhesives were determined. The failure load obtained from the failure criterion maximum shear strain predicts the experimental failure load with a mere difference of 2% for the MS-polymer Soudaseal 270 HS. Due to the stress singularities in the adhesive layer with the 2cepoxy 3M™ Scotch-Weld™ 9323 B/A the "stress at a distance" approach must be applied for this adhesive. Only the failure criterion maximum shear stress with the stress considered at a distance equal to the full adhesive thickness from the stress singularity predicts the experimental failure load. These failure criteria predict the experimental failure loads of the local models conservatively and accurately. For adhesives between flexible adhesives and stiff adhesives, both failure criteria must be applied, the lowest failure load will be the actual failure load.