Random Fatigue Test Sampling Requirements
This citation is provided as a resource for researchers, but Contrails cannot provide a full-text download
U.S. government employees, Military/Department of Defense employees, and U.S. government contractors and sub-contractors may be eligible to register with the Defense Technical Information Center (DTIC), where this report and others like it may be available
Report Number: AFFDL TR 65-95
Author(s): Choi, S. C., Enochson, L. D.
Corporate Author(s): Measurement Analysis Corporation
Laboratory: Air Force Flight Dynamics Laboratory
Date of Publication: 1965-08
Pages: 76
Contract: AF 33(615)-1314
DoD Project: 4437
DoD Task: 443706
Identifier: AD0622086
Abstract:
Basic considerations are discussed for determining sample sizes and record lengths for various statistical tests and estimates which are important to random fatigue testing. Methods for determining minimum sample sizes when comparing means and variances of normally (Gaussian) distributed random variables are described. Procedures for reducing a relatively large sample to a smaller sample are presented. Elimination of outliers and systematic resampling are two methods given. An explanation is presented of the requirements and problems involved in the determination of record lengths necessary for an estimate of a given accuracy for autocorrelation functions, ordinary power spectral density functions, cross-correlation functions, cross-spectral density functions, frequency response functions, and probability density functions. Due to its importance in random fatigue testing applications, the basic properties of the Weibull distribution in terms of its parameters and the failure rate are summarized. A presentation is given of estimation and statistical testing problems related to the Weibull distribution. The best available methods of estimating the parameters are described. Methods of determining sample sizes needed for various analyses are developed. Some problems of reliability analysis applicable in fatigue testing are discussed. New methods of decision techniques for comparing two or more systems are proposed in terms of reliability. The report concludes with an example of the application of the Weibull distribution to actual fatigue test data.
Author(s): Choi, S. C., Enochson, L. D.
Corporate Author(s): Measurement Analysis Corporation
Laboratory: Air Force Flight Dynamics Laboratory
Date of Publication: 1965-08
Pages: 76
Contract: AF 33(615)-1314
DoD Project: 4437
DoD Task: 443706
Identifier: AD0622086
Abstract:
Basic considerations are discussed for determining sample sizes and record lengths for various statistical tests and estimates which are important to random fatigue testing. Methods for determining minimum sample sizes when comparing means and variances of normally (Gaussian) distributed random variables are described. Procedures for reducing a relatively large sample to a smaller sample are presented. Elimination of outliers and systematic resampling are two methods given. An explanation is presented of the requirements and problems involved in the determination of record lengths necessary for an estimate of a given accuracy for autocorrelation functions, ordinary power spectral density functions, cross-correlation functions, cross-spectral density functions, frequency response functions, and probability density functions. Due to its importance in random fatigue testing applications, the basic properties of the Weibull distribution in terms of its parameters and the failure rate are summarized. A presentation is given of estimation and statistical testing problems related to the Weibull distribution. The best available methods of estimating the parameters are described. Methods of determining sample sizes needed for various analyses are developed. Some problems of reliability analysis applicable in fatigue testing are discussed. New methods of decision techniques for comparing two or more systems are proposed in terms of reliability. The report concludes with an example of the application of the Weibull distribution to actual fatigue test data.