Engineering Software Research and Development, Inc. USA
Dr. Barna Szabó is co-founder and chairman of Engineering Software Research and Development, Inc. a company whose mission is to create and market software tools for the advancement of the quality, reliability and timeliness of information that serves engineering decision-making processes.
Dr. Szabó was a full-time member of the faculty of the School of Applied Science and Engineering at Washington University in St. Louis from 1968 until his retirement as the Albert P. and Blanche Y. Greensfelder Professor of Mechanics in 2006. His areas of expertise include mathematical modeling techniques, methods for the assurance of the reliability of engineering decisions based on computed information.
Dr. Szabó is the principal author of two textbooks on finite element analysis (John Wiley & Sons, 1991 and 2011) and has published over two hundred technical papers, mostly on the finite element method. Dr. Szabó is an external member of the Hungarian Academy of Sciences, Fellow of the St. Louis Academy of Sciences, holds an honorary doctorate and is a founding member and fellow of the US Association for Computational Mechanics.
On the Formulation and Application of Design Rules
Numerical simulation in connection with the design and maintenance of mechanical and structural components subjected to cyclic loads is addressed from the perspectives of designers and authorities responsible for the formulation of design rules.
Design rules are typically stated in the form: Fmax ≤ Fall.That is, the maximum value of some functional F must not be greater than its allowable value Fall. Since Fmaxis computed numerically, only an approximation to Fmax, denoted by Fnum, is known. Therefore it is essential to have means to estimate the error Fmax- Fnum otherwise the design cannot be certified. Furthermore, numerical error penalizes design. This underlines the importance of solution verification, a fundamental technical requirement.
The formulation of design rules involves the application of verification, validation and uncertainty quantification procedures. The goal is to establish rules for design that make it extremely unlikely that failure would occur during the service life a component. This involves measurements of fatigue life using test coupons. Statistical models are used for the generalization of test data.
The stress in the test section of the coupons is either constant or varies smoothly. Therefore it is necessary to formulate predictors suitable for generalization of the test results to arbitrary stress fields. Many plausible generalizations are possible. A new class of predictors and their calibration and ranking will be described and illustrated by examples. It is shown that ranking of predictors is possible only when experiments are performed outside of the range of calibration.
Since design rules are conditioned on the experimental information available when the rules are formulated, revision of design rules is necessary when new information becomes available. Proper and efficient application of design rules, the management of empirical data accumulated over time and systematic updating of mathematical models used in the formulation of design rules require an organizational support structure and exercise of simulation governance.