Simulation modeling and analyses of straight horizontally-laminated timber beams
Abstract
Reliability Based Design Methods are being adopted for many materials in structural applications. One of the materials that is yet to use this design method is wood. Recent developments, however, indicate a renewed enthusiasm in the wood industry along this line. One of the wood products that will require a tremendous amount of research to implement the Reliability Based Design Method is the laminated beam. Implementation of this method requires strength and stiffness properties of laminated beams in terms of population distributions. To perform the large amount of testing for a large product such as the laminated beam would be prohibitive in terms of cost. A viable alternative would be the development of suitable computer programs to predict the properties as accurately as possible. This thesis is an example of such an effort. A stochastic model capable of predicting the properties of straight horizontally laminated beams was developed. The model is based on the well established and much used deterministic model known as the "I$\sb{\rm K}$/I$\sb{\rm G}$" model. A careful study of this model revealed certain features that indicated the possibility of transforming them into equivalent probabilistic features. Such a transformation was performed. Two systems were developed. One uses the transformed section method of analysis and the other, the finite element method of analysis. Because a complete stochastic analysis requires thousands of repetitions, the systems were implemented in the Cyber 205 super computer at Purdue University. The use of suitable vector operations enabled the creation of a very efficient finite element code. An efficient vectorized substructuring technique incorporated in the finite element code can be used to estimate the properties of large beams with reduced central processing time and page faults. The "I$\sb{\rm K}$/I$\sb{\rm G}$" model seems to predict strength values that are in good agreement with the actual strength data made available by the American Institute of Timber Construction. The model, however, seems to underpredict the stiffness data by about 10 percent. As an added feature, a stochastic glulam model that was developed for the American Institute of Timber Construction was incorporated in the systems. This model underpredicted the strength and stiffness data by about 50 and 10 percent, respectively. For both the models, the transformed section method of analysis seems to produce results that are almost identical to those produced by the finite element method of analysis.
Degree
Ph.D.
Advisors
Meyers, Purdue University.
Subject Area
Civil engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.