Verification versus validation

We have used the words "verification" and "validation" associated with concepts of model accuracy and quality without stating the precise meaning. There is, however, no universal understanding about the meaning of these two words [195, 196]. To avoid questions about terminology, it is important to have precise and unambiguous definitions. Throughout this thesis, we will use the following definitions consistent with recommendations made in 1979 by the S.C.S. Technical Committee on Model Credibility [254]. It must be noted that in some application areas the definitions of verification and validation are, unfortunately, interchanged [196].

Definition 4.1 (Model verification) The process of determining whether or not a computer simulation model is consistent with the underlying mathematical model to a specified accuracy level.

Verification is often coupled to the adjective "internal" to stress that the main goal of verification is to check the implementation of the model equations in the timedomain simulation software (or wind turbine design code).

The first stage of verification is concerned with checking that the structure of the simulation program is consistent with the underlying mathematical model (e.g. by comparing the simulated steady-state values with those determined analytically from the mathematical model or visually inspect the correctness of the responses). The second stage of the verification is concerned with numerical accuracy. Simulation of mathematical models generally involves the integration of sets of ordinary differential equations. The performance in terms of accuracy and speed varies for different models and settings (including step size and tolerance). To select the most suited integration algorithm (e.g. Runge-Kutta fourth order (RK-45), Gear or Linsim), the different algorithms can be compared at the same relative error per integration step. Furthermore, in the case of fixed step integration methods, comparisons can be made of results with a number of different integration step sizes.

Once the checks of the structure of the program has been completed satisfactory, and no algorithmic problems have been identified, the next step is to validate the verified model. Model validation is defined as:

Definition 4.2 (Model validation) The process of determining whether or not the verified mathematical model of a system behaves similar to the real behavior associated with the intended model use.

That is, validation is the process of proving that the verified model is an accurate approximation of the real system under investigation. Thereto, the model behavior has to be compared with real-life properties to check if the assumptions upon which it is based are satisfied. Validation is, as opposed to verification, closely linked to the adjective "external" to emphasize the connection with the real system under investigation. It is important to recognise that the validity of a model is closely linked to the intended use. That is, a model developed for one purpose may not be appropriate for another.

Solar Stirling Engine Basics Explained

Solar Stirling Engine Basics Explained

The solar Stirling engine is progressively becoming a viable alternative to solar panels for its higher efficiency. Stirling engines might be the best way to harvest the power provided by the sun. This is an easy-to-understand explanation of how Stirling engines work, the different types, and why they are more efficient than steam engines.

Get My Free Ebook


Post a comment