Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Engineering Science and Mechanics


This study addresses the growing need for appropriate methods of analysis and simulation of service loads. Specifically, the study attempts to provide a method for analyzing nonstationary service loads that uses available tools of analysis, requires a modest computational facility, and reduces the results to a few parameters. The study also addresses the problem of finding a simulation method that uses these analysis parameters to provide a reproducable service load;A new presentation of the state of art of the methods of analysis and simulation of service loads is given;A nonstationary model is presented that represents the nonstationary process as a composition of two different stationary processes. These stationary processes are assembled according to a probabilistic model (generalized Poisson process) to form the nonstationary process. The idea of parameterizing the description of service loads is presented;In the analysis phase, the method of segmenting the nonstationary sequence and a statistic that estimates each segment population are used to obtain power spectrum estimates of the stationary populations which constitute the nonstationary signal. The method of smoothed periodograms was used as the computational technique of power spectrum estimation. Two different population estimators were used. A method for estimating the statistical parameters of the generalized Poisson process is given. Finally, the coefficients of two fourth order digital filters were used to describe the power spectra of the stationary processes;In the simulation phase of the proposed method, software and hardware methods are presented to generate a white random sequence of numbers, generate the generalized Poisson process, shape the white sequence into a sequence with the required power spectra and finally generate the nonstationary sequence;The proposed method of analysis and simulation of service loads was applied to a typical analog record and a typical digitized sequence of data;The proposed method is seen to be successful in providing a practical way of analyzing a nonstationary signal, presenting the analysis results in terms of few parameters, and generating a nonstationary sequence at a fast sampling rate that can be used by engineers for fatigue life prediction programs or fatigue life testing of components and structures.



Digital Repository @ Iowa State University,

Copyright Owner

Mohamed Khalid Abdelhamid



Proquest ID


File Format


File Size

165 pages