Summary and Conclusions

Numerical parameter studies based on FEA have been performed for structural bonding designs exposed to constant surface loads that are adjusted to the bond design stresses of ETAG 002. Special focus was placed on concentrated corner loads as predicted by thin plate theory. The FEAs confirmed the theoretical result for the case of SSG models that the effect of corner loads significantly influences the stress distribution in the corners. This effect leads to averaged stresses in the corner area that can be higher than nominal design values according to ETAG 002. Thus, the stress levels induced by the corner loads can­not be neglected in the assessment of the overall mechanical performance of SSG.

TABLE 2—Averaged stress levels versus peak stress levels for selected configurations; surface load of 4.48 x 10~3 MPa (glass unit: 1.5 m in length, 1.25 m in width, varying thickness; bonding: width = 20 mm, thickness = 6.7mm).

Glass unit configuration

Averaged corner stress, MPa

FEA corner stress, MPa

8 mm



10 mm



2 x 6 mm without shear



2 x 6 mm without shear



These averaged corner stresses increase for the following:

• increasing aspect ratios of the glass panes,

• increasing glass bending flexibility, and

• increasing bonding stiffness.

Nevertheless, successful real world installations for more than two decades proved the applicability of the ETAG 002 recommendations for SSG designs within the limited scope of ETAG 002. Removing structural silicone from the corners or reducing its cross-section at the concerns in order to let the corners flex during wind events (recall also Fig. 6 in this context) might seem to be a the­oretical solution, but it is not applicable in the real world because of air and water infiltration issues.

Unfortunately, several design parameters of the bond geometry suitable for reducing corner loads directly counteract favorable design solutions for the improved durability of cold bending warped designs. In general, cold bending techniques favor a small glass bending stiffness and large bites in order to keep the permanent cold bending stresses low. Due to low design stress limits for per­manent loads, cold bending designs are highly sensitive to changes in the vari­ous key performance parameters. Therefore, the often-made request to select for low glass stiffness (if possible) and large bites is evident. However, this com­bination is detrimental for the occurrence of corner loading. As the aspect ratio is given by design considerations, one key parameter remains open for specifica­tion: the height of the bond. For low corner loads induced by pressure loading, a large height is favorable. However, the warping requirement—or, more gener­ally, the cold bending requirement—itself might pose an upper limit for the flex­ibility of the SSG from architectural point of view. Due to these contradictory requirements, bond geometry tailoring (e. g., dedicated bond height or width distributions along the edges) will probably be only partially effective, if at all.

A potential geometric parameter for improved stress distribution and thus durability might be the warping field itself. In the analysis, a bi-linear warping field is assumed for simplicity. In reality, warping of the glass units might be achieved by built-in shapes (warping established by design), elastic bending (warping by elastic bending), or inelastic bending (warping based on plastic hinges) of the supporting frames. This paper does not answer the question of to what extent the stress distribution evoked by warping depends on the different warping fields for a given corner offset. The related optimization problem neces­sitates a warping design that improves the stress distribution in the corners due to warping and still fulfills architectural and cost efficiency requirements.

Alternative approaches might involve the investigation of non-rectangular glass units with respect to their suitability for cold bending or cylindrical or conical bending patterns. These proposals significantly impact the architectural design space.

ETAG 002 gives the impression of vanishing corner loads in the case of sur­face loads such as wind, etc. Thus, the reader might suspect that the problem of mastering the wind loads is almost independent of the consideration of the cold bending permanent loads. As has been shown in this paper, the related stress peaks associated with surface loads and cold bending loads are highly coherent in space. Thus, the interference between these load cases is obvious and should

Copyright by ASTM Int’l (all rights reserved); Tue May 6 12:07:08 EDT 2014

Downloaded/printed by

Rochester Institute Of Technology pursuant to License Agreement. No further reproductions authorized.

be quantified in more detail for the durability assessment of facades featuring cold bending assembly techniques. Please note that within the framework of ETAG 002, permanent shear design stresses are 1 order of magnitude lower than the shear design limits for unsteady loads. For tensile design stresses, no values are specified for permanent loads, but a reasonable extrapolation from shear to tensile loading involves a similar reduction of the stress limits by 1 order of magnitude.

From a mechanical point of view, the true performance of SSG adhesives under the related operating conditions—either creeping under permanent loads or creeping plus unsteady loads—and thus the technological limits of SSG, are not really known. The development of an improved knowledge of the material performance attributes of the adhesive that might result in reduced safety mar­gins might improve the applicability of cold bending techniques significantly. In this context, it has to be kept in mind that cold bending failures might be attrib­uted to different issues such as air and water infiltration and aesthetic deficien­cies (or glass breakage). Thus, different requirements are related to the mechanical performance of the bond and the adhesive, and these need to be covered by adequate technological know-how and design procedures.

From an approval point of view, a simple and straightforward extrapolation of ETAG 002 does not seem possible due to the simplifying assumptions regard­ing the stress distribution of the bond. One way of extending the applicability of ETAG 002 is to parameterize corner stresses and warping stresses in a similar fashion. Nevertheless, in order to be conservative, the simpler the assumptions and the procedures, the higher the required safety margins (and/or the more limited the scope), with adverse impact on the available design space for cold bending. It is highly likely that dedicated FEA is better suited for such kinds of applications, allowing one to achieve optimum design solutions with respect to today’s knowledge and the technological limits. Parametric finite element stud­ies accompanied by dedicated tests for fundamental baseline cases might be used to compile the results for standard configurations into graphs or tables that could be used for pre-design purposes.