Peruvian Standards Differ from U.S.

The Wiese Group, owners of Banco Wiese, created the largest financial institution in Peru upon its 1998 merger with an Italian financial group. Before the merger, however, the Wiese family built a national headquarters in Lima and enlisted the help of Galvez and Sheen to verify the durability of the building’s decorative glass and aluminum curtain wall.

The Vistawall Group of Terrell, Texas, prefabricated the wall in the United States and FAM Peru was the Peruvian contractor. But building codes are different between the U.S. and Peru, where standards allow greater building movement under seismic events.

Such movement is called story drift, and is defined as the relative displacement of a building slab over the floor height beneath it. In American code, allowable story drift for a building with the specifications of Banco Wiese is 0.005. In Peru, allowable drift for the same building is 0.007.

Story drift would cause the curtain wall to move at points where it was attached to the building with an aluminum frame. Between the frame and the outer glass is a layer of silicone, which deforms with the frame and keeps the glass from shattering or disconnecting from the frame and falling.

In the end, Sheen and Galvez used Algor software to prove that the wall would withstand seismic activity as great as 7 on the Richter scale.

Determining a Safe Shift

Because the curtain wall chosen for the Banco Wiese building was designed in America to withstand story drift of 0.005, Galvez and Sheen wanted to see if the extra 0.002 of potential story drift allowed by Peruvian standards would cause the wall to fail.

Galvez and Sheen figured the distance between slabs at the floor separation levels (360 cm) and multiplied that by the allowable story drift (0.007) to determine a prescribed displacement for their analysis.

"For the Wiese Building, the distance between slabs is 360 cm," Sheen said. "The story drift limitation for that kind of building is 0.007, so we expected a 0.007x360=2.52 cm displacement – around an inch." They used prescribed displacement to account for that motion in their analysis.

Using Superdraw III, Algor’s precision finite element model-building tool, Sheen used thousands of brick elements to create more than 10 models of different parts of the wall. The models represented the building’s three distinct sides – a vertical wall, a sloped wall and a circular wall. Each model was tested for in-plane and out-of-plane movement.

Sheen selected three material models from Algor’s material model library to build his models. He chose a linear model to represent the aluminum frame of the wall. To model the silicone that connected the aluminum to the glass, Sheen picked a Mooney-Rivlin material model. For the glass outer surface, he used a Von Mises model.

When defining the material properties of his Mooney-Rivlin material, Sheen remembered training he got at Algor headquarters in Pittsburgh. It was there that he learned he could determine certain constants for Mooney-Rivlin behavior from stress/strain curves. He contacted the silicone’s American manufacturer. Based on the manufacturer’s stress/strain information and the way that data matched up with Algor’s available material models, Mooney-Rivlin proved to be the best material model for Sheen’s analysis.

Sheen found material properties for the glass model at the Glass Association of North America and used common engineering resources to define aluminum’s linear material properties.

Sheen chose Algor’s Mechanical Event Simulation technology- invented by Algor to provide a virtual laboratory and eliminate the need to input dynamic loads by determining the motion, flexing and resulting stresses of a part or assembly at each instant of an event - because he wanted to simulate on his nonlinear models the dynamic loads the building would be subjected to in the event of an earthquake. He wanted to take advantage of Algor’s contact elements to check the behavior of the wall’s support system, namely the bolts that anchor the wall to the building.

Simulating Seismic Influences

The model built, Sheen performed a linear static stress analysis to verify the model’s geometry. Sheen and Galvez agreed that it was important to make sure their model had no errors in it before running the Mechanical Event Simulation.

For the linear static stress analysis, Sheen applied wind loads to the model because comprehensive studies of wind loads meant there was data against which he could check his results. For boundary conditions, Sheen fixed the lower points of the model in the spots where the curtain wall would be attached to the building. The analysis showed Sheen that his model was geometrically sound.

He then applied a prescribed displacement to the model of 2.52 cm (about 1 inch) for his MES. Galvez and Sheen found some points of stress concentration in the corners of their model. The findings were consistent with other studies of curtain walls, Sheen said. The stress was acceptable, mainly because the load condition, a 7-magnitude earthquake, was a once-in-a-lifetime event.

When the analysis was complete, Galvez and Sheen used Algor’s postprocessing tools to make animated reproductions of their findings. Galvez and Sheen presented their work with Algor to peers around their country, including last year’s National Civil Engineering Congress.