Glass Belis, Bos & Louter (Eds.) Conference on Architectural and Structural Applications, Ghent University, September 2020. Copyright © Author. all rights reserved. ISBN 978-94-6366-296-3, https://doi.org/10.7480/cgc.7.4765

Erin Mills Town Centre is a shopping mall in Mississauga, Canada, owned by Cushman & Wakefield in Toronto. The revitalization project was designed by MMC International Architects Ltd., with a glass steel sphere with a diameter of 27.4m in its central court. The structural glass of this spherical structure is made of laminated hyperboloid glass panels supported by a steel structure.

Due to the thermal stress generated between the glass frit on the glass panel and the internal and external lites of the laminated component, some challenges emerged during the construction process. Testing is used to establish a viable solution for replacing the glass to correct in-situ problems found.

Cushman & Wakefield hopes to revitalize the existing Erin Mills Town Center shopping mall in Mississauga, Ontario, Canada. The architectural design team led by Chris Brown of MMC International Architects Ltd. challenged our office to create a unique glass and steel feature in the central court of the mall. Several options were provided for ownership review, and a structural globe was selected.

The globe glass patch is supported on a series of steel rings (similar to the latitude lines on the earth), which are in turn supported by seven random steel rings. The earth structure is also built with an inclination angle of 23.5 degrees to represent the angle of the earth's axis. The glass panel of this project is composed of three layers of hyperboloid glass. Coat the #2 surface of the laminate with ceramic frit, and coat the #6 surface with a hard coat low-emissivity coating, in an effort to provide some comfort for the occupants in summer conditions.

The initial design proposed point-supported glass panels and fully tempered or thermally strengthened panels. It turns out that, as described in this article, this is difficult to achieve, so it was decided to switch to SMD accessories with annealed glass panels. The design load and stress levels are low enough to allow this type of structure, however, due to the thermal stress generated by the structure's geometry, which creates unexpected loading conditions, corrective measures are needed.

Figure 1 below shows the steel structure supporting the spherical glass panels. We used seven random rings that intersect each other to create a support structure for the evenly spaced rings to form the shape of the earth and support the glass panels. The seven rings support the entire structure as a whole and are supported on the brackets on the reinforced roof structure (see Figure 2), forming the opening of the center court. All-steel is an exoskeleton structure on which glass panels are suspended, creating a very smooth glass surface inside the building space. This is done to help promote the flow of air along the inner glass surface and to keep this movement as evenly distributed as possible.

The glass panels are laminated to improve strength and safety. Since the panel is integral, an important mechanical system is installed at the bottom of the spherical structure, and blowing nozzles are installed at the bottom of the entire structure to ensure good air flow and help cool the space and glass in summer. The glass is designed with low-emissivity coatings and ceramic frit patterns to help control solar heat gain. The original concept was to use point-supported spider connectors to suspend the glass panels on the steel structure. This will require tempered glass and some creative double-curvature glass manufacturing.

As mentioned earlier, the first glass component consists of tempered glass. Initially, the design required three layers of 6 mm glass interlayer 1.52 mm SentryGlas interlayer. This design concept evolved into 5 mm all-tempered lite/10 mm chemically-tempered lite/5 mm all-tempered lite. The panel was constructed by collapsing 10 mm thick glass into an appropriate hyperboloid shape and then chemically tempering it. This creates a shape for the outer and inner 5mm fully tempered flat lite to be laminated and held in place with a strong interlayer. Accelerated testing under high humidity to see if the lamination can be maintained throughout the life cycle of the glass component.

In addition to assembly tests and adhesion tests, load/strength tests were carried out in the steel and light metal structure laboratories of the University of Munich and the Faculty of Applied Sciences in Germany. Figures 3 and 4 below show some of the test procedures done on larger panels.

Both the strength test and the accelerated adhesion test performed well, but we did encounter problems with assembly consistency and visual defects. Figures 5 and 6 below show these delamination issues (especially after a piece of lite is broken), and the inherent optical anisotropy/visual distortion in the process.

Due to the problems found in the load test and the model, it was decided that if all three lites are thermally collapsed annealed glass, and the outer lite still has frit on the surface #2, then the component will look better, the inner lite on the surface# 6 has a hard coat low-emissivity coating. This means that the fastening system is modified from a point-supported spider connector to a patch-mounted spider connector. Turning to annealed glass, the assembly was modified to three identical 8 mm thick lites.

The manufacturing and installation of the glass panels are based on the strength and stress calculation results of the new FE model. Due to the shadow created by the exoskeleton structure on the glass, the thermal load was modeled and found to be within the allowable range. However, during the construction process, the contractor (Josef Gartner GmbH) noticed cracks in several panels. After examining the fracture mode of the broken glass, it was determined that most of the failures were caused by excessive thermal stress in the external sintered glass. In the analysis of the revised components, the fused glass has reduced the strength of the glass (by 40%), so further investigation is needed to determine the cause.

After careful inspection, it was found that all thermal stress cracks originated from the glass frit point. As shown in the figure below, this is consistent across all cracked panels. We are beginning to understand where the problem is, but we need a way to find an appropriate solution that we can trust. At this time RJC hired Dr. Jens Schneider to review the problem at hand and design a test that would allow us to determine the cause and then recommend a solution to the problem.

Information collection is implemented first. The on-site readings of the internal and external glass surface temperature are taken during the summer months so that we can determine the thermal gradient of the laminated component from the inside to the outside. Take readings at several locations every 15 minutes.

The temperature reading results show that the difference between the indoor ambient temperature and outdoor ambient temperature in summer is within 5°C, but the difference in surface temperature is significant. Surface #1 (outside) is in the range of 35°C, while surface #6 (inside) is in the range of 55°C. This creates a temperature difference of approximately 20°C through the cross-section of the laminated glass panel under the worst conditions.

Collect an in-situ glass sample by cutting out a part where the thermal crack started (see Figure 9) to gather more information. These were cataloged and brought back to the laboratory of the Darmstadt University of Technology.

The microscopic image below shows the frit taken at the edge of the glass (including the chamfered edge). Obviously, this is a problem and causes the component to fail prematurely. In addition, specimens collected from the field were opened at the crack line to check the fracture mode and evaluate the potential cause. These inspections also confirmed that the problem was directly at the frit position on the edge of the glass panel.

The next step is to obtain samples of the glass used in the laminated glass assembly for further strength testing and see how the manufactured glass compares with the allowable stress levels we used in the design simulation. Four-point bending test and double-loop bending test are used to determine these strength levels.

The test samples are divided into five different series. The four-point bending test is used for the first four series of specimens, and the double loop test is used for the fifth series. These are summarized in the table below.

Series 1-3 are test samples cut from the edges of three different glass panels commonly used in laminated components. The only difference with these samples is that their edges are polished instead of the edges that were polished according to the field panels previously provided to the site.

Series 4 is a test sample cut from the edge of a sintered glass panel, where the edge of the frit is removed a distance from the edge of the panel. The actual samples in this series do not have any glass frit on the cut samples, and have been edged.

Series 5 is a sample with glass frit on the entire surface. The sample was cut from the center of the glass panel and used for surface stress testing.

A finite element model was also performed to understand the stress level that we might expect from the gradient temperature difference across the thickness of the glass component. This is based on in-situ monitoring at a thermal gradient of 20 °C. Based on the results of these sample tests and the finite element model, it is determined that the glass component should be able to withstand climatic loads and thermal stress loads, provided that the glass frit is far from the edge of the glass panel and the gradient is 20° C or less. It is also noted that the glass strength of the panel edge where the frit extends to the edge is actually closer to 40% of the non-sintered glass value, compared to the 60% assumed in the original calculation.

In order to further satisfy that the new glass components removed using the frit edge 10 cm from the edge of the panel can withstand the expected thermal stress, additional tests were performed on the full-scale sample to see what thermal gradient would cause stress failure in the panel. Panel (see Figures 11 and 12). This test is used to confirm that the thermal gradient expected on site will fall within the safe area of ​​the actual capacity of the glass assembly.

The test was performed using three different components. Series 1 is based on the in-situ panel (glass frit on the edge of the glass). The missing edge of series 2 is 10 cm, while series 3 has no frit at all. For series 1 specimens, the failure probability curve produced by the test is shown in Figure 13 below. Series 2 and 3 cannot be drawn in a similar way, because the glass sample is only broken by the in-plane thermal stress caused by the test setup, and will not be broken by the thermal gradient between surface #1 and surface #6.

Through all the tests performed and the finite element model, the following conclusions were drawn:

Read Jones Christoffersen Ltd. thanks the ownership (Cushman & Wakefield) for their cooperation in designing a unique glass structure for the Erin Mills downtown project. We would also like to thank Joseph Gartner for his expertise and help, as well as Dr. Jens Schneider and the Darmstadt University of Technology team for all their guidance to ensure the success of the project.

3196 Thompson Road Fenton, MI 48430 United States

Polígono Industrial El Bayo, parela I, 19 24492 Cubillos del Sil León Spain

9551 Ray Lawson Avenue. Montreal QC H1J1L5 Canada

Via Del Lavoro Snc 64023 Mosciano Sant'Angelo TE Italy

Ismail Horse. D-100 Karayolu CAD. No:44A, 34947 Tuzla, /İstanbul Turkey

Via Marcatutto 7, (Milano) Italy 20080 Albairate MI Italy

9361 Irvine Avenue Irvine, CA 92618 United States

Via Fieramonti 1 04012 Cisterna di Latina LT Italy

19, rue du Puits Romain, L-8070 8070 Bertrange Luxembourg

Via del Lavoro, 1 22036 Erba CO Italy

No. 160, Yichuan Road, Jiaonan City, Qingdao City, Shandong Province, China 266000

Boll. Ind. Penapurreira Parcela C4-B, 15320 As Pontes de García Rodríguez A Coruña Spain

1616 W. Williams Dr. Phoenix,, AZ 85027 United States

carat. Km from Estacion. 15.8 44415 Rubielos de Mora Teruel Spain

No. 20, Binhe Road, High-tech Development Zone, Luolong District, Luoyang City, Henan Province 471003

Industrijska zona RZ R29 51223 Kukuljanovo Croatia

2200 W. Salzburg Road Auburn, MI 48611 United States

Carrer del Pla, 108-110, Pol. Ind. El Pla 08980 Sant Feliu de Llobregat Barcelona Spain

400 Guys Run Rd Cheswick, PA 15024 United States

Runów, ul. Solidarności 1 05-504 Złotokłos Poland

Building 8, SIRIM Complex, No. 1, Persiaran Dato' Menteri, Section 2, PO Box 7035, Darul Ehsan, 40700 Shah Alam Selangor Malaysia

779 Washington Street Buffalo, NY 14203 United States

Via Molare 23/c 15076 Ovada AL Italy

Log in or register to post a comment