Main /
WP 3 And WP 6Stefan Ledin (14.03.2006) Patrick Roache devoted a chapter to the semantics of verification and validation, Roache (1998). Roache's and the AIAA's definitions are very similar. Roache simplified the definition ever further with Verification : mathematics Validation : science/engineering Roache said that a code cannot be validated, but only calculation(s) of specific type of problems can be validated. This is a point made by Vladimir at the WP3 and WP6 meetings in Oslo. Reference: Roache, P. J., Verification and Validation of Computational Science and Engineering, Hermosa Publishers, Albuquerque, New Mexico, USA, 1998. Dmitriy Makarov (27.2.2006) Definitions from AIAA G-O77-1998 "Guide for the Verification and Validation of Computational Fluid Dynamics Simulations" you may find interesting: "Verification provides evidence that the model is solved right. Verification does not address whether the model has any relationship to the real world. Verification activities only evaluate whether the CFD model, the mathematical and computer code ... is solved accurately." "Validation, on the other hand, provides evidence that the right model is solved" ... "Validation is the process of assessing the degree to which the computational simulation represents the real world" Thomas Jordan (24.2.2006) Some basic assumptions: existing (applicable) CFD software to
Actually I don't see validation potential in these applications (I admit this statement is strict, but it relies on the very demanding meaning of the word "validation", see below) WP6 - CFD Club + benchmarking is mainly about research in and further development of (sub-)models, increase the understanding of the actual abilities. I expect an systematic assessment framework to be set up, which could help to assist even experienced users to select models and to chose the best set of parameters for most reliable CFD results. I would like to fix here our wording. So please enter your perception on the words "In general, validation is the process of checking if something satisfies a certain criterion. Examples would be: checking if a statement is true, if an appliance works as intended, if a computer system is secure, or if computer data is compliant with an open standard. This should not be confused with verification."
However, in my understanding validation implies that the CFD results in all possible cases should be valid, i.e. in a small scatter band compared to the "true" reference results. Is validation possible on a restricted set of examples? Or does one have to mention immediately the example or set of examples against which the validation has been done? Who does define the reference solution ("standard" above) and in the second step determines the "compliance". "In the context of hardware and software systems, formal verification is the act of proving or disproving the correctness of a system with respect to a certain formal specification or property, using formal methods. As opposed to formal verification, Software testing cannot prove that a system does not have a certain defect and it cannot prove either that it does have a certain property. Only the process of formal verification can prove that a system does not have a certain defect or does have a certain property. Either way it is impossible to prove or test that a system has "no defect" since it is impossible to formally specify what "no defect" means. All we can do is prove that a system does not have any of the defects that we can think of, and has all of the properties that together make it functional and useful." "In the context of hardware and software systems, formal verification is the act of proving or disproving the correctness of a system with respect to a certain formal specification or property, using formal methods.
As opposed to formal verification, Software testing cannot prove that a system does not have a certain defect and it cannot prove either that it does have a certain property. Only the process of formal verification can prove that a system does not have a certain defect or does have a certain property. Either way it is impossible to prove or test that a system has "no defect" since it is impossible to formally specify what "no defect" means. All we can do is prove that a system does not have any of the defects that we can think of, and has all of the properties that together make it functional and useful." |