Numerical treatment of engineering problems with uncertainties. The fuzzy set approach and its application to the heat exchanger problem

Abstract

An important part of predicting responses with computational models is the definition of systems parameters such as material properties. The reliability of model-based predictions depends on the quality of these data, and the knowledge of their uncertainties. When experiments cannot be performed to determine parameter values, they must be obtained from the literature, namely handbooks, standards, specification sheets, etc. This paper addresses the use of literature-based parameters and other information for assessing uncertainty in computational predictions. The uncertainties in the data and the solution (prediction) are described in the framework of fuzzy sets. A double-pipe heat exchanger thermal analysis problem is used to illustrate the difficulties with computational predictions when only informations from the literature are available. It is shown that the resulting uncertainties are significant and cannot be neglected. Copyright © 2010 John Wiley & Sons, Ltd.