FASTENER ANALYSIS CORRELATES CLOSELY WITH PHYSICAL TEST

Gene O'Fallon, Hilti Inc. product development engineer.

The analytical determination of stress for a complex threaded fastener has predicted failure loads that correlate almost perfectly with tests on a prototype. As a result of successful verification for a single design, a manufacturer has been able to apply Algor engineering software to analyze a series of design variations entirely by analytical means.

Hilti Inc., a manufacturer of construction fastening systems, performed the study on a female adhesive anchor designed to secure structural members to concrete base material. Most government and commercial construction, for example, use adhesive anchors to fasten steel to concrete. The primary objective of the study was to evaluate and compare the mechanical designs of two different female anchors by finite element analysis (FEA) and by experimental methods.

"Basically, we needed to study a couple of internally threaded anchor designs that had different external thread geometries," explains Gene O'Fallon, product development engineer for Hilti.

A female adhesive anchor.

"The challenging part of analyzing this anchor is accounting for the position of the internal and external threads relative to each other. This cannot be predicted for a manufactured part. If an internal and external thread root happen to be opposite each other, a local thin spot in the anchor wall is created. This thin spot is a potential failure site if the location coincides with a highly stressed region of the anchor. In lieu of extensive and costly experimentation, the only effective way to study this problem was to use the finite element method and construct several models.

"Due to the complexity of the model, we were not anticipating good agreement between the FEA and experimental results," O'Fallon recalls. However, in terms of yielding, the prototype shoed the anchor to yield at 17,500 - 23,00 pounds load, whereas Algor's SuperSAP showed the anchor to yield at 18,000 - 20,000 pounds load. "Needless to say, we were delighted. The models showed peak von Mises stress in the same area the actual anchors failed. The SuperSAP results also agreed well with our experimental load-strain data."

Five different axisymmetric finite element models were constructed and analyzed. The model geometries varied in external thread design and relative internal and external thread locations. On the average, each model consisted of 2500 2-D elastic and 64 truss elements. The bolt, anchor, surrounding adhesive, and concrete comprised each model. Analysis was done overnight on an IBM PS/2 computer.

Interactive Means Easy

"SuperDraw II made detailed modeling and meshing easy, and I really liked the boundary condition features," O'Fallon notes. "First, I constructed models of the anchor, adhesive, and concrete. It was easy for me to define the properties of adhesive, concrete, or steel for various model parts by assigning different group numbers to the lines that made up elements in the drawing. During decoding, the SD22SS program allowed me to input new property values for each group number used in SuperDraw II. These properties were then assigned to the appropriate elements in the model. With 2500 elements, this was a real time saver.

"Following this, I created a model of the bolt. Using Substruct, the model of the anchor and bolt were 'glued' together such that the mating threads did not touch. Then I used AEdit to add very stiff truss elements between the threads to connect the two models, thus creating a single model of the anchor system. Each anchor model was run using SSAP0. After the first run, I eliminated those trusses that went into tension and reran the models in an iterative manner. Only three iterations were required for convergence. This iterative solution method allowed me to see where the load was being taken."

Parts of the adhesive anchor finite element model defined.

O'Fallon was most impressed with Algor's high performance but low price. "I liked the comprehensive package and its powerful modeling and meshing techniques. The attention Algor paid to making the software highly interactive with clear messages at each step made the system both user friendly and easy to understand."

The displaced model and stresses were evaluated using SuperView and POST. "SuperView's features were a plus in analyzing the results efficiently and effectively. The speed of color dithering was excellent. With SuperView's post analysis stresses, we were able to determine which anchor design would yield first with respect to relative internal and external thread geometries and orientation.

The deflected anchor after analysis.

"I am familiar with other finite element programs, and the yield theory criteria that SuperSAP offers were very helpful, O'Fallon continued. We used von Mises theory to analyze the anchors themselves, since it is the most accurate in comparison to experimental results for ductile materials. We also used the Maximum Principal Stress theory to study the adhesive and concrete, since they are more brittle in nature."

From the analysis, O'Fallon was also able to determine the following:

• How the anchor deflects under load,
• Magnitude and location of peak stresses,
• Contribution of base material to anchor holding,
• Effect of friction in the threads, and
• Load each anchor design could support before yielding.

With the aid of a color inkjet printer, some exciting and interesting color stress contours were created as well.

Copyright ?1989 Algor, Inc. All rights reserved.