#### Installation — business terrible - 1 part

September 8th, 2015

Under high steady stresses, materials may undergo time-dependent deformation resulting in failure called creep rupture which limits their lifetime. Structural silicones were first subjected to this type of testing by Sandberg and Rintala [34]. The original testing was done using tensile/adhesion joints (H-pi – ece specimens) that used a block of structural silicone sealant cast between parallel plates that measured 12.5 mm x 12.5 mm x 50.8 mm. These dimensions were in accordance with the dimensions stated in ASTM C1135-00 (2011) Standard Test Method for Determining Tensile Adhesion Properties of Structural Sealants [35]. In the current study, a similar test protocol was used to study the dead load resistance of structural silicone film adhesive bonded between stainless steel buttons and glass substrate.

Specimens for testing the creep resistance of point fixing on glass were prepared by bonding a 20 mm diameter stainless steel “button" with a threaded socket head (see Fig. 3) to a glass plate (cross section of adhesive interface: 314 mm2). The structural silicone film adhesive (1 mm initial and 0.8 mm final thickness) was cured between the button and the glass by placing the complete assembly into an industrial autoclave operated at a pressure of 1.275 MPa and a temperature of 130°C for a total of 25 min.

Creep rupture testing was performed at ambient laboratory climate conditions (23 ± 2°C, 50 ± 5% relative humidity) by loading the stainless steel button fixations in tensile with weights of 20 and 40 kg, corresponding to dead loads of

1.25 and 0.63 MPa, respectively. The specimens exposed to 1.25 MPa load failed, on average, after 7 years, while no failures were observed for the specimens loaded with 0.63 MPa after now more than 11 years (as the time of this writing). Figure 7 shows the tensile loading creep rupture experiments conducted in the laboratory.

A separate experiment was set up to evaluate the behavior of the bonded steel button/glass specimens described above in shear loading by monitoring the time to failure. Again, testing was carried out at ambient laboratory climate.

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.

FIG. 7—Tensile loading creep rupture experiments conducted in the laboratory. (Note weights on the leftare 20 kg and 40 kg loading a 20 mm diameter steel button in tension, with an adhesive bond area of 314 mm2.) |

Figure 8 shows the experimental setup and the specimen orientation chosen in order to place the load in perfect shear mode.

When pulled in an extensometer at a rate of 6 mm/min, the five specimens failed cohesively within the TSSA layer at an average maximum shear stress of

4.25 MPa. When loaded with a constant shear load of 3.40 MPa, all five specimens failed within a few seconds. The average time to failure was estimated to be around 1.4 s (future tests will utilize an electric mechanism in order to record the time to failure more exactly). When loaded with a constant shear stress of 2.55 MPa, the five replicates failed between 5 and 6 h, with an estimated average time to failure of about 5.5 h. At a constant shear stress of 1.95 MPa, the five specimens failed within 4 to 24 h, with an estimated average time to failure of about 14 h. The five replicates subjected to a shear load of 1.70 MPa failed after

FIG. 8—Experimental setup and specimen configuration for testing time to failure in shear mode. 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. |

34, 49, 49, 113, and 126 days. All specimens that failed during the experiment thus far failed in cohesive failure mode within the TSSA layer. As of the time of this writing (190 days after starting the test procedure), no failures have been observed under loads of 1.42, 1.13, 0.85, and 0.57 MPa. Figure 9 shows the average time to failure values obtained at different shear stress levels on a set of five replicate test specimens.

As can be seen from Fig. 9 with the data plotted on a log/log scale, the logarithm of the time-to-failure periods shows an apparent linear relationship to the logarithm of the constant shear stress levels applied in the creep rupture experiment. The linear appearance on a log-log plot corresponds to a power relation of the form

rcreep — A ‘ tfai2 (1)

For polymeric materials it has been shown that double logarithmic plots of stress versus failure time often yield straight lines [36]. This allows the use of the power law shown in Eq 1 to fit the data with a least-squares regression line, which minimizes log creep stress errors. Using the average time to failure data, the best fit is obtained with the A — 2.1611 and B — -0.064, with shear stress expressed in MPa units and time in days. This least square fit is associated with a coefficient of determination R2 of 0.9269. Using the shortest times to failure experienced at all loads yields a power law fit with A — 2.0844 and B — -0.067 and a R2 of 0.8737. Using the longest time to failure seen at all loads thus far into the experiment yields a fit with A — 2.2059 and B — -0.061 and a R2 of

0. 955. At the failure load observed in the extensometer testing (4.25 MPa), these power laws give time to failure periods between 1.85 and 2.22 s.

Obviously these power laws can also be used to extrapolate future failure events. Given the current, limited amount of data and the uncertainty in the

FIG. 9—Average time to failure obtained at different constant shear stress levels on a set of five replicate test specimens. 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. |

accuracy of the short time failure periods, such extrapolations are speculative in nature. However, it is still useful to predict the time to failure periods for the next lowest shear load levels that are applied within the current range-finding experiment. Use of the above power law trend curves yields time to failure predictions for the 1.42 MPa shear load of 732 days (average), 318 days (shortest), and 1420 days (longest). If these “ball park" figures are only approximately correct, then a new series of experiments with higher shear loads needs to be designed, as the least square fit to the shortest failure times observed thus far does not yield any failure within a 30 years time period for loads of 1.118 MPa or less. Obviously, such extrapolation of short-term creep rupture data assumes that no change in the creep or degradation mechanism occurs over the predicted service, an assumption that is less likely to hold true the more one tries to extrapolate into the future (for example, the power law extrapolations used above still give a residual strength 10 times greater than the currently accepted dead load design strength value for structural silicone glazing sealants after >1010 years exposure duration). However, the concept is still valid; testing at higher sustained loads (approaching the bond’s ultimate strength as determined in the extensometer testing) necessarily results in failure after short periods of time, while lower levels of load provide corresponding longer time periods prior to failure. As shown in Fig. 10, such a curve would show, at some level of sustained load, an asymptotic “run-out" behavior, whereby failure does not occur for any reasonably anticipated time duration.

The tensile rupture tests discussed previously provide further corroboration that the above extrapolations may hold true within reasonable timeframes. The failure for the specimens loaded with 1.25 MPa in tensile that actually occurred, on average, after 7 years, is predicted to occur, based on the shear creep rupture test data extrapolations, to occur between 5.7 (minimum) and 30.3 (maximum) years, with an average time to failure predicted of 14.2 years.

As can be seen from Fig. 10, at some point in time, the safety margin between the sustained load curve and the permissible design load becomes constant. This approach is outlined in ASTM D 4680 Test Methods for Establishing Allowable Mechanical Properties of Wood-Bonding Adhesives for Design of Structural Joints [38] and its merits discussed in a review of durability test methods and standards [39]. However, significantly more work will be required (and is planned) to demonstrate such behavior under different accelerated weathering regimes for the TSSA film adhesive.

A further consideration in deriving the dead load design strength of the TSSA material is what happens if the point supported glass pane fails. When glass is supported by point connections, typically no more than two points are required to support the dead weight. Figure 11 shows a drawing of a glass pane with six attachment points noting the top two points are required to support the dead load. Additional connection points will be used to support the live load only. This is due to the reality of construction tolerances. Therefore the weight of a piece of glazing supported by two attachment points will have to be part of the design, and furthermore a single attachment point must support the dead weight of the entire glass panel in the event of a mechanical failure of the hardware or glass defects. Whatever the permissible load is determined to be, it

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

Downloaded/printed by

FIG. 11—Schematic of a glass pane with six attachment points noting the top two also take the dead load of the glass. |

must provide sufficient short term duration at double its value to allow for the repair and replacement glazing.

For example, if the permissible load is determined to be 0.6 MPa for constant stress, then the adhesive must have integrity at 1.2 MPa to allow for fix and repair. Noting Fig. 6 above, 16 weeks of constant loading in tension at

1.25 MPa demonstrated only a minor change in ultimate strength when pulled to destruction in tension. Shear loading at 1.42 MPa has resulted in no loss of bond after 180 days. Such data applied to the reality of the application give assurance to the determination of the design load. For example, a value of

0. 6 MPa can very well be a reasonable dead load design stress for TSSA when the endurance limit is confirmed. On the other hand, if the constant permissible dead load design stress is determined to be 1.0 MPa based on endurance limit validation, the fact that dead load shear at 1.95 MPa resulted in material failure in one day, provides an uncomfortable situation if the adhesive is required to sustain a constant load in the event one attachment point is damaged during the service life. The derivation of the static design strength must consider and reflect the reality of field repair and serviceability.

Furthermore, one must also consider the fact that while a point support is subjected to a constant shear load due to the dead weight of the glazing, a live load due to wind events will subject the adhesive to tension loads normal to the plane of the glazing (out-of-plane loads). The combination of tension and shear loading on specific attachment points needs careful evaluation. Sandberg and Ahlborn [40] confirmed through testing of structural silicone materials that the interaction between nominal tensile and shear forces is elliptical based on the following equation

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

Downloaded/printed by

where Fs and Fl are the ultimate strengths in shear and tension, respectively, fs and ft are the actual stresses under test. The data that was taken on the structural silicones showed that the ultimate shear stress was roughly equal to the tension stress. This is also the case with the TSSA material presented in this paper.

When combining shear design stress and tension design stress for TSSA, the following equation should apply

JL

F2 + F2d

s des t des

where fs and ft are the actual stresses in dead load shear and live load tension and Fs des and Ft des are the permissible design loads in dead load shear and live load tension, respectively. This will ensure that there is enough strength in each attachment point to support both the long term dead weight and live loading without exceeding the permissible stresses on the point.