Mechanical connections in cold-formed steel: comprehensive test procedures and evaluation methods

Design equations do not exist for connections in cold-formed steel formed with most types of mechanical fasteners, and such designs are presently based on test results. A lack of standard test procedures and corresponding evaluation methods impedes the acceptance of connection designs and the exchange of information vital to understanding the connection performance of a large number of mechanical fasteners with diverse properties. Initial steps toward standardization are taken with the development of fixtures and procedures for single shear, pullout and pull-over tests. The proposed single shear tests employ both a simple specimen formed with two overlapping straps and a more sophisticated fixture designed to simulate the behavior of a connection in a cold-formed steel shear diaphragm. A single test fixture applicable to any specimen shape and stiffness is developed for pull-out and pull-over tests. Test results, including results of large scale pull-over tests, obtained in the development and verification process are compared and discussed. Twenty-five identical tests on typical specimens are performed in accordance with each of the proposed test procedures to provide data for the determination of probability distributions on ultimate strength. A beta probability law is assumed as the underlying model for the probability density function on connection strength, and Bayes\u27 theorem is used with a diffuse prior to evaluate the joint posterior likelihood function on the beta parameters in digital form. Marginal density functions on the parameters are determined and the Bayesian distribution on connection strength is evaluated. Samples of various size are drawn from the distribution by simulation, and probability density functions on the sample mean are determined by a method of moments fit to a beta model. The Bayesian distributions are compared with distributions obtained with maximum likelihood estimates of the beta parameters to ascertain the suitability of the computationally efficient maximum likelihood approach for model determination with limited sample sizes. The test evaluation method is based on a load and resistance factor design criterion with a first order probabilistic code format. An expression is developed for the coefficient of variation of the resistance which includes the effects of strength variation, the size of the test sample and the degree of simulation of actual field conditions. Resistance factors are determined for connections in temporary, standard and vital structures made with different levels of workmanship and inspection. It is found that three tests are sufficient to estimate the mean resistance and that a single set of resistance factors is appropriate for use with the proposed test procedures