(294a) Validation of Models and Procedures in Chemical Engineering
AIChE Annual Meeting
2015
2015 AIChE Annual Meeting Proceedings
Liaison Functions
IACChE's James Y. Oldshue Lecture
Tuesday, November 10, 2015 - 8:30am to 9:30am
In recent years, the term “validation” has become extremely widespread among researchers working in the worldwide chemical engineering community. It is applied to a wide range of models and procedures in an effort to convince the reader that the work and the approach adopted are of value, as well as to justify extensive description of the model details and continued effort. However, the word “validation” should convey the meaning of establishing not just value, but validity, based on objective criteria which include testing over a wide range of variable space over which the model is intended to be applied. It should require agreement within objective criteria based, among other factors, on experimental errors, where the experimentation has been conducted independently and, wherever possible, based on different techniques and carried out by several independent research groups. Parameter fitting, where applicable, should not have been based on the data used to test the degree of agreement.
Widespread misuse has rendered the terms “validation” and “validated” almost meaningless. At a time when chemical engineers are, more than ever, needed to provide input and perspective on key global issues like climate change, water resources and food resources, it is important that we have objective ways of assessing the validity of models and other tools. Based on his experience as both a modeler and an experimentalist working in several different areas of chemical engineering, several questions will be considered: 1. When, if ever, is it justified and permissible to say that a model has been validated? 2. Are qualifiers needed when the words validation and validated are used? 3. What terminology should be applied in cases where validation cannot be established? 4. How do we persuade graduate students and their supervisors to exercise more care and rigor in applying these terms?