Discussion of the Finite-Element Modeling and Analysis

To predict the behavior of the mockup designed for this study before the actual testing, a finite-element model was developed and analyzed. The mockup detail that is discussed subsequently is shown in Fig. 7. The finite-element model was constructed in a program called Visual Analysis by Integrated Engineering Soft­ware [17]. This finite-element modeling program focuses primarily on linear – elastic modeling of structures. Figure 8 shows an overall view of the completed model. The aluminum mullions (vertical and horizontal) are represented by 2-node beam elements. Because of the flexibility of the silicone sealant at the pe­rimeter of the glass lites, the glass panel is expected to act as shear-resisting ele­ments, and, therefore, the moment restraints were released where the horizontal mullion beam elements attach to the vertical mullion beam elements. It is noted that the stiffening influence of a nominal moment connection between the aluminum mullions is expected to be insignificant compared to the stiffening influence of the glass lites, which are attached to the aluminum

TABLE 4—Properties of weathered sealants under tensile testing.

Sealant

Ultimate tensile stress psi (kPa)

Ultimate tensile strain (%)

Young’s modulus psi (kPa)

Dow Corning 995 RT

162(1117)

210

231(1593)

Dow Corning 995 RT + 1 h 88° C

160(1103)

146

236(1627)

Dow Corning 995 RT + 1 h -29° C

162(1117)

216

230(1586)

Dow Corning 995 RT + 5000-h UV exposure

145 (1000)

171

200(1379)

Dow Corning 983 SGS RT

165 (1138)

116

486(3351)

Dow Corning 983 SGS RT + 1 h 88°C

146(1007)

82

485 (3344)

Dow Corning 983 SGS RT + 1 h -29°C

151 (1041)

128

364 (2510)

Dow Corning 983 SGS RT + 5000-h UV exposure

170(1172)

156

401 (2765)

FIG. 7—Mockup designed for preliminary testing.

mullions by the silicone sealant. The structural silicone sealant and the glass lites are represented by 3- or 4-node plate elements. The appropriate linear – elastic material properties for all three materials were entered into the model. For the silicone sealant, the anisotropic and non-linear material behaviors can­not be accounted for in this specific finite-element modeling software. The Young’s modulus, E, of the silicone sealant was varied in separate runs of the model in an attempt to capture both the tension and shear behaviors of the seal­ant. This was only a rudimentary attempt to capture their effects. A more

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. 8—Finite-element overall model view. (a) mullion beam elements with end releases, and (b) glass and sealant plate elements.

sophisticated finite-element software program would be needed to more fully integrate these non-linear sealant properties into the analysis.

For the actual testing, movements in the mockup specimens were measured at discrete displacement values. When trying to obtain analytical results from computer software that can be appropriately compared to testing results, it is important to simulate as closely as possible the actual test conditions. There­fore, the finite-element model was displaced to the same displacement values that were measured in testing. Figure 9(a) shows a graphical representation of the displaced shape of the finite-element model at a given displacement. Whereas the actual cyclic testing was carried out at frequencies of 0.8 Hz and 0.4 Hz, respectively, for displacements less than or equal to 3 in. (76.2 mm) and larger than 3 in. (76.2 mm), the finite-element analysis was carried out stati­cally, which simulated a very slow loading rate. A close-up of one of the glass lite intersections can be seen in Fig. 9(b), where the movement of the panel below with respect to the upper panel at the horizontal stack joint is identified by the vertical line break at the bottom nodes of the upper panel. In both of these figures, the glass is represented by a “blue" color and the silicone sealant by a “grey" color.

The first step in evaluating the finite-element model is to review the overall results and determine if they generally correlate with the testing results and

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. 9—Finite-element model displaced shapes: (a) overall displaced shape, and (b) close-up of displaced shape.

what might be considered to be appropriate behavior. From the overall view of the displaced model (Fig. 9(a)), it is evident that the plate elements, representing the glass lites, are displacing and rotating rigidly. The close-up view of the glass lite intersection (Fig. 9(b)) also reveals that the silicone sealant is being stretched in an elastic manner. The finite-element model clearly demonstrates similar movement behavior to that of the actual test. Because the Young’s mod­ulus, E, of the glass is 10100 ksi (69640 MPa), and the E of the silicone sealant varies between 45 psi (310 kPa) for shear and 400 psi (2758 kPa) for tension (Tables 2 and 3), i. e., substantially more flexible than the glass lites, this type of result would be expected. The beam elements representing the aluminum mul – lions are not visible in either of the finite-element model views represented by Fig. 9. Because the strong and weak axis rotational degrees of freedom at each end of these elements are released (no moment), there is no resistance to lateral displacement from these elements in the finite-element model. Comparison of the finite-element analysis result with mockup test results is described subsequently.