Introduction

There is a saying in engineering that ‘when things start breaking, usually multiple things have gone wrong.’ In the case of fritted spandrel glass, we are seeing the traditional go-to-solution of heat strengthened glass and ceramic frit exhibiting thermal fracture failure. Many things have changed since this was the common successful solution. Much investigation has gone into the strength resistance aspect of the equation, however there also have been changes on the thermal stress demand placed on our spandrel glass units.

Historically ceramic frit included lead and other materials which are now excluded from material production. Changes in chemistry mean that the frit of today is not the same as the frit that has been used successfully for so many years. (Barry et al. 2015)

The spandrel glass units of 20 or 30 years ago were often single glaze with reflective coating on surface 1, later reflective on surface 2 with an opacifier or a ceramic frit on surface #2 (Barry et al 2015). In order to achieve aesthetic uniformity, as well as improved thermal efficiency, spandrel glass units now are commonly double glazed with double or triple silver coatings on surface #2 and the ceramic frit on surface #4 (Barry et al. 2015). Additionally, the code requirements for improved energy performance means that spandrel panels are now commonly highly insulated, with four to six inches of insulation (or more) and possibly include mullion wraps. The configuration of our spandrels is now much more effective at collecting heat and retaining it. The thermal stress that the glass units are subject to is much greater; consequently, we need to look not only at the available strength of the glass, but also at the level of demand that the glass is being subjected to. In the past, with single glaze spandrels, the question was between annealed (AN) and heat strengthened (HS) glass; historically, ASTM E2431 Determining the Resistance of Single Glazed Annealed Architectural Flat Glass to Thermal Loadings has served that purpose well, but as the design paradigms change, so do the checks that need to be performed. Now the question is whether glass used should be Heat Strengthened (HS) or Fully Tempered (FT).

ASTM has a standard for thermal stress on single glazed units (E2431) but does not yet have a standard for double glazed units. In the absence of an American standard, the French standard NF DTU 39 has been used extensively in Europe and is getting increased usage in the United States. This standard includes a strength reduction factor for fritted glass and predicts that the modern spandrel glass units with HS glass will be overstressed in modern spandrel glass and shadow box applications.

This paper presents some of the available data, example analysis cases in a moderate climate, alternative products, and potential design strategies. It also seeks to highlight implications for glass design in general.

Glass Strength

The United States and Canada conduct glass design based on the Beason and Morgan Glass Failure Probability Model (GFPM) with the implementation of a stress-probability integral over the area of the glass (E1300, CAN/CGSB 12.20). Much of the rest of the world is based on principal tensile stress models of some sort or another, such as EN16612, DIN 18008 or CEN/TS 19100 in Europe or AS1288 in Australasia.

Changes of Design Environment

Using French Standard NF DTU 39 Part 3 as a basis (it is only available in French, but Google translate does a reasonable job for those with some familiarity in the area) we can see how the change in design practices is affecting the demand on our spandrel glass.

NF DTU 39 Part 3 includes strength reduction for FT with enamel, but does not clarify the reduction for HS glass, so HS without frit is used as a common basis for comparison of changing spandrel design conditions unless noted otherwise in the description.

Table 1: Working stresses per NF DTU 39 Part 3 Table 10. Source: NF DTU 39 Part 3 (translated)

Evaluations are for Seattle, WA (not an extreme environment) using WinTHS software; 50mm (2in) airspace behind the glass and 2in or 4in insulation, a low thermal mass frame (best case option), and vertical glass with unpolished edges, subject to partial shading. NF DTU 39 uses a series of 1 dimensional calculations for the center of glass, edge zone and frame to calculate the thermal differentials.

Table 2: Utilization of Inner Lite in Seattle (WA USA) with different spandrel configurations

Seattle is not a particularly extreme environment. There are areas of the United States that have extreme cold combined with low-angle sun and clear skies that would be more severe. In the table above, sixteen percent (16%) overstress is probably still within the realms of safety factor of “working stresses” and the bracketing of frames into “low” “medium” and “high” thermal inertia without full 2D/3D analysis. However, if we note below that the characteristic strength of HS with Frit (45 MPa) is the same as AN glass (45 MPa) then a more serious situation is apparent with utilization of 212.6% overstress. Typical HS levels of Residual Compressive Surface Strength (RCSS) are rarely at the minimum. The fractures that have been observed with fritted HS glass is a warning that we are close to the ultimate limit with limited safety factor, not just for HS fritted glass but also for HS in modern spandrels and shadow boxes using IGUs with LowE coatings combined with high levels of insulation.

Figure 1 Thermal profile in spandrel as analyzed per NF DTU 39.

Strength of Fritted Glass

NF DTU 39 includes strength reduction for frit on FT glass but does not provide a figure for HS with enamel (frit). CEN/TS19100-1 Design of glass structures - Part 1: Basis of design and materials (CEN/TS 19100) quotes reductions for strength for both HS and FT glass.

Table 3: Source: Table 5.3 CEN-TS 19100-1 Design of glass structures - Part 1: Basis of design and materials

Similar reductions are also confirmed in bending tests by Chris Barry and Scott Norville however it should be noted that results vary widely depending on the type of frit, color (hence composition), coefficient of thermal expansion, and pattern.

Four-Point Bending Test – Ceramic Frit Results (ASTM C1161-13)

Figure 2a Four-Point Bending Test, Full Distribution Heat Strengthened Glass

Figure 2b Four-Point Bending Test, Detail for 0.008 Probability of Failure Heat Strengthened Glass

Figure 3a Four-Point Bending Test, Full Distribution Fully Tempered Glass

Figure 3b Four-Point Bending Test, Detail for 0.008 Probability of Failure Fully Tempered Glass

Source Figures 2 and 3: Barry, C.J., Norville, H.S., Unexpected Breakage in Ceramic Enameled (Frit) HS IG Spandrels Paper for IGMA Winter Conference, Fort Lauderdale, Florida, 2015

There are three outcomes of note from these experimental results:

• The HS-frit cumulative probability is almost vertical,
• The HS-frit line is well to left of the HS (no frit) line, and
• The HS-frit is less than no-frit at the 8/1000 probability of breakage level.

The greatest significance of the verticality of the line with frit is that it does not match the distribution assumed in the Beason Glass Failure Probability Model (GFPM) which assumes failure will occur at a location based on a flaw distribution, rather than at the point of peak stress. Without the GFPM being valid, ASTM E1300 is substantially not applicable. The implication of a near vertical line is that there is a uniform flaw distribution and that failure will occur at the point of highest stress.

In particular, the “Analytic Method” in Annex A2 of ASTM E1300-16 should be treated with extreme caution, as some testing indicating adequate capacity by other methods in E1300 (See Bergers, M., Natividad, K., Morse,S.M., Norville, H.S., Full Scale Tests of Heat Strengthened Glass with Ceramic Frit) has recorded failures at less than the predicted usable capacity when the actual as-tested RCSS is used in conjunction with Annex A2. In Lingnell, A.W; Beason, Dr W.L.; Brackin, Dr M.S. ASTM E1300 Uniform Load Strength Reduction Factor not Required for Ceramic Enameled Glass, conclude that the more traditional methods are adequate for fritted glass, however that conclusion was made based on glass with elevated RCSS; a better conclusion may be that with elevated RCSS, E1300 excluding Annex A2 may be applicable. Given the uniform frit/flaw density, principal stress analysis, limiting stresses to those in section X6, which limits HS glass to 46.6MPa (6,750 psi) and 93.1MPa (13,500psi) may be more appropriate for fritted glass, however verification with the glass/frit manufacturer is recommended.

The recently published version of ASTM E1300 includes a note, "Ceramic enamel is known to affect glass load resistance.
Consult the manufacturer for guidance." The relationship between RCSS and bending strength is apparently broken, so the process in ASTM C1048 Condition B, which measures RCSS optically after removal of the frit to calculate a strength, is unlikely to be representative of the available strength. Arguably, removal of the frit renders the sample unusable. Destructive testing by 4-point bend tests to ASTM C1161, ring-on-ring tests (ASTM C1499 modified to scale for thickness of glass) or methods in ISO 1288-3 are better to determine actual available strength by direct application of stress rather than by implication from RCSS. This is particularly valid since the flaw distribution is near uniform and the failure distribution is near vertical.

The actual process of weakening the glass surface is not fully known to the author, however in the micrograph below we see that the frit intrudes into the surface of the glass and seals it from moisture, preventing self-healing.

Figure 4 Frit in glass surface flaws. Source: Bergers, M., Natividad, K., Morse,S.M., Norville, H.S., Full Scale Tests of Heat Strengthened Glass with Ceramic Frit, Challenging Glass, 2016

Fracture Location

Figure 5 In situ fracture Source: Barry, C., Norville, H.S., Unexpected Breakage in Ceramic Enameled (Frit) HS IG Spandrels, 2015

Figure 6 In situ fracture Source: R.Green, Confidential Project

Figure 7 In situ fracture Origin is away from edge Source: R.Green, Confidential Project

Looking at the above 3 photographs, each origin of fracture is not at the edge of glass, as is typical for single glaze systems; rather it corresponds to the line of the spacer bar. In the lower photo we see a large extension of the glass edge beyond the structural silicone to framing, so while in some cases cold bridge from the framing (such as at stack joints) may participate, the cold-bridge of the IGU spacer appears to be a more significant influence. Where the spandrel box is hot, the exterior is relatively cold and the spacer bar is the most conductive connection between the two conditions, the selection of spacer bar will control the heat flux, the local temperature in the glass, and the thermal stress gradient.

Strategies for Spandrels

Increase the minimum RCSS

The Residual Compressive Surface Strength (RCSS) is a function of the heat strengthening process. Both HS and FT glass have similar procedures with different cooling rates. Heat Strengthened glass has an RCSS range from 24 MPa to 52 MPa (3500 to 7500 psi), which is in addition to the inherent working tensile strength of about 23.3 MPa (3380 psi) away from the edge. It is possible that the reason we have yet to see more wide-spread problems is because glass fabricators tend to target the middle of the range, not the minimum. Specifying an increased minimum will have the same effect, however this is a strategy that some fabricators will agree to and some will not. Further increasing RCSS and using FT glass is also an option. Provided the glass with the as-fabricated RCSS meets the minimum anticipated bending strength by testing to ISO 1288-3 and achieves strength similar to the table from CEN/TS 19100 for unfritted glass, then similar overall strength can be anticipated.

Note that in the higher ranges of RCSS for HS glass, spontaneous fracture due to Nickel Sulfide (NiS) is possible, but as the inner lite is critical and is contained on both sides it has limited life-safety risk. Frequency of NiS fracture in HS is lower than FT; many consider the commercial risk in HS to be acceptable. Indeed, some go as far as to say NiS fracture is negligible or non-existent. The lower inherent tensile stress in the glass requires a much larger and statistically rarer NiS stone to fracture HS glass. (See also NiS in HS Glass, Kasper et al.) For both HS and FT glass, heat soak testing (per ISO 20657) will significantly reduce the risk of NiS failure in service.

Figure 8 Examples of NiS Stones Source: NiS in HS Glass, A Technical Paper: Dr A. Kasper; J. Colvin; F Serruys https://www.saint-gobain-glass.com/nis-in-heat-strengthened-glass

Thermal Breaks and Warm Edge Spacers

Stresses at the observed fracture locations are driven by a combination of the frame and the IGU spacer; the incorporation of thermal breaks in the frame and warm edge spacers in the IGU reduce the temperature difference at the edge of glass. In particular, these strategies reduce the stress line at the spacer where failures have been observed. The reduction in strength due to frit may focus concern on the loss of strength, however it should be a warning for use of HS glass in general. NF DTU 39 indicates we are changing the environment our glass is subjected to, thus we should be looking carefully at the edge detailing and materials.

Silicone Opacifiers

Testing has shown that silicone opacifiers do not decrease the strength of the glass and, in the case of ball drop tests, may in fact increase the available strength. Noting that there is limited tail at the lower end, this may be as simple as the opacifier protects the surface from weathering and accumulation of flaws, so acts more like “fresh” glass.

Figure 9a Comparison of glass coated with silicone opacifier and uncoated glass - heat strengthened

Figure 9b Detail for 0.008 Probability of Failure Comparison of glass coated with silicone opacifier and uncoated HS glass

Figure 10a Comparison of glass coated with silicone opacifier and uncoated glass - Fully Tempered

Figure 10b Detail for 0.008 Probability of Failure Comparison of glass coated with silicone opacifier and uncoated FT glass

Figure 11 Ball Drop height comparing uncoated, silicone and ceramic fritted FT glass.

Source Figures 9,10,11: Vockler, K.L, Krytenberg, T.P., Norville, H.S., Blanchet, S., Swanson, J.W., Barry, C.J., Carbary, L.D., Hoffman, S.P., Torok, G.R., and Fronsoe, C.S., Silicone Opacifiers for Spandrel Glass Applications: Risk Mitigation in Thermal Stress, Presentation at Glass Performance Days 2017, Tampere, Finland, 2017

Conclusions

Changes in design practices to increase the thermal efficiency of building envelopes is increasing the thermal stress on IGU spandrel glass. Studies in a moderate climate such as Seattle, and failures in more sever climates indicate that additional analysis and changes and design practices are warranted.

The failures observed in fritted Heat Strengthened (HS) glass in over 30 large-scale projects is a warning for the industry in general, both for the degree that spandrels are becoming critical for thermal stress, and in implications for strength of fritted glass and the capacity to support wind and other loads.

Testing of fritted glass has shown a distinctly different failure distribution to that used in the Glass Failure Prediction Model. Caution should be exercised using ASTM E1300 with fritted glass and, in particular, caution using the ‘Analytic Method’ in Annex A2 of ASTM E1300-16. Peak principal stress per Appendix X6 may be more appropriate in a context in which the cumulative probability distribution is near vertical. Future editions of ASTM E1300 will include a note to confirm the strength with the glass supplier.

Thermal analysis of the spandrel area to NF DTU39 is a good starting point but lacks the specifics to capture detailed 2D effects, such as spacer type and detailed frame effects at the edge of glass.

Silicone opacifiers do not decrease the glass strength; rather, in some cases they may increase the strength of glass thus significantly improve the capacity of glass to resist thermal fracture and other loads relative to fritted glass.

Use of thermally broken frames and warm edge spacers reduce the cold bridge at the edge of glass and reduce the risk of thermal fracture.

Thanks to Chris Barry and Casey Anderson who provided figures and permission to reprint.

References

NF DTU 39 P3 Norme Francaise: Building works — Glazing and mirror-glass works — Part 3: Calculation memorandum for thermal stress. Published and distributed by the French Standardization Association (AFNOR)

Anderson C., A Risk Mitigation Strategy for Thermal Stress Breakage - Vitrum Glass Talks, Seattle, WA USA 2023

ASTM C1048 Standard Specification for Heat-Strengthened and Fully Tempered Flat Glass

ASTM C1161 Standard Test Method for Flexural Strength of Advanced Ceramics at Ambient Temperature

ASTM C1499 Standard Test Method for Monotonic Equibiaxial Flexural Strength of Advanced Ceramics at Ambient Temperature

ASTM E1300 Standard Practice for Determining Load Resistance of Glass in Buildings

Barry, C.J., Norville, H.S., 2015 Unexpected Breakage in Ceramic Enameled (Frit) HS IG Spandrels Paper for IGMA Winter Conference, Fort Lauderdale, Florida,

Barry, C.J. 2023 Enameled Spandrel Up-Date or Problem Solved!

Bergers, M., Natividad, K., Morse,S.M., Norville, H.S., Full Scale Tests of Heat Strengthened Glass with Ceramic Frit, Challenging Glass, 2016

CAN/CGSB-12.20-M89. Structural Design of Glass for Buildings. Standards Council of Canada

CEN/TS19100-1 Design of glass structures - Part 1: Basis of design and materials CEN/TC 250 European Committee for Standardization

Chrysanthi A. - Thermal Breakage of Glass - Comparison and Validation of thermal shock calculation methods. Master Thesis Project, Delft University of Technology, 2016

DIN 18008-1 2020 Glass in Building – Design and construction rules – Part 1: Terms and general bases

Elstner M, Polakova M, Schafer S, Thermal Stress Analysis in Glass, Façade Techtonics, 2020

ISO 1288-3 Glass in building — Determination of the bending strength of glass — Part 3: Test with specimen supported at two points (four point bending)

ISO 20657:2017(en) Glass in building — Heat soaked tempered soda lime silicate safety glass

Kasper, Dr A.; Colvin, J. ; Serruys F.: NiS in HS Glass A Technical Paper: https://www.saint-gobain-glass.com/nis-in-heat-strengthened-glass

Lingnell, A.W; Beason, Dr W.L.; Brackin, Dr M.S. ASTM E1300 Uniform Load Strength Reduction Factor not Required for Ceramic Enameled Glass. GPD Glass Performance Days, 2017 https://www.gpd.fi/GPD2017_proceedings_book/

Schwind G, Paschke F, Schneider J Case Studies on the Thermally Induced Stresses in Insulating Glass Units via Numerical Calculation - Challenging Glass Conference Proceedings – Volume 8 – 2022

Vockler, K.L, Krytenberg, T.P., Norville, H.S., Blanchet, S., Swanson, J.W., Barry, C.J., Carbary, L.D., Hoffman, S.P., Torok, G.R., and Fronsoe, C.S., Silicone Opacifiers for Spandrel Glass Applications: Risk Mitigation in Thermal Stress, Presentation at Glass Performance Days 2017, Tampere, Finland, 2017