My question is about the subject of the validation process of the results of a numerical study, for example, via Ansys fluent. There is something I noticed in several articles on this topic and I did not understand it. I hope someone who has an explanation or an explanation for that would be kind enough to write it. The topic, in brief, is the numerically studied model, for example, a plate heat exchanger with new plates. To perform the validation process of the results of the numerical study, the results are compared with another numerical or experimental study of a plate heat exchanger that uses traditional chevron plates published in the scientific literature! I know that any change in the calculation grid changes the accuracy of the results, so the matter will be if the model is completely changed. Does anyone have an explanation or explanation for my understanding of this?
with thanks and respect
Hello Ahmad Aboulkhail ,
There are two processes used to check modelling; 1) verification, and 2) validation.
In verification, you are asking did you use the model right? This is a mechanical process of checking that you used the model correctly, i.e, that you did not make any errors in applying the model. There are many errors you can commit such as error in the values of material parameters, faulting boundary and/or initial conditions, improper grid spacing, etc.
In validation, you are asking did you use the right model? This is a more involved process because you are asking if the model really is a valid simulacrum of the physical process. For example, suppose you have a thermal diffusion equation. This equation can be applied to many different problems with different geometries, different shapes, different materials, different boundary conditions, different initial conditions, etc. As long as the different systems satisfy the assumptions of the underlying model, then the model should accurately predict the behavior of the different systems. In the case you mentioned, the two types of heat exchangers, you need to examine the underlying model(s) that were used in these two different cases. It may be the same model, and as long as the two different heat exchangers satisfy the assumptions of the model, then the results should be correct for both systems. The difficultly is determining which model or models was or were used with each numerical simulation. If the same model was used, and the results were correct in both cases, then this could be thought of as a partial validation of the model in use. Do you know the underlying model(s) for the two different heat exchangers; are they the same underlying model? The model(s) is (are) usually some type of differential equation.
Regards,
Tom Cuff
Hello Ahmad Aboulkhail ,
There are two processes used to check modelling; 1) verification, and 2) validation.
In verification, you are asking did you use the model right? This is a mechanical process of checking that you used the model correctly, i.e, that you did not make any errors in applying the model. There are many errors you can commit such as error in the values of material parameters, faulting boundary and/or initial conditions, improper grid spacing, etc.
In validation, you are asking did you use the right model? This is a more involved process because you are asking if the model really is a valid simulacrum of the physical process. For example, suppose you have a thermal diffusion equation. This equation can be applied to many different problems with different geometries, different shapes, different materials, different boundary conditions, different initial conditions, etc. As long as the different systems satisfy the assumptions of the underlying model, then the model should accurately predict the behavior of the different systems. In the case you mentioned, the two types of heat exchangers, you need to examine the underlying model(s) that were used in these two different cases. It may be the same model, and as long as the two different heat exchangers satisfy the assumptions of the model, then the results should be correct for both systems. The difficultly is determining which model or models was or were used with each numerical simulation. If the same model was used, and the results were correct in both cases, then this could be thought of as a partial validation of the model in use. Do you know the underlying model(s) for the two different heat exchangers; are they the same underlying model? The model(s) is (are) usually some type of differential equation.
Regards,
Tom Cuff
I'm afraid my question was misunderstood.
My question, in particular, is whether the numerical modeling of an engineering model (with Ansys fluent)(a plate heat exchanger with an innovative plate) can be validated by comparing the results with the results of an experimental study of another engineering model (a plate heat exchanger with a plate different from the innovative plate). And if that is permissible, what are the conditions that must be met for the comparison to be true?
Hayder I. Mohammed
Thomas Cuff
For the comparison to be true, it is necessary to compare with results under the same initial and boundary conditions as your results.
Good luck@
For a true validation, the modelled can be compare to the experimental results using established statistical techniques or programmable tools.
A distinction must be made between theory and model. A theory is an ideal or simplifiable situation that can be rigorously verified by experience. For example, Coulomb's law (in electrostatics) or Newton's second law (in classical mechanics) are rigorously verified by experience. This is not a validation of a model.
A model is a complicated situation with often a large number of parameters and equations. In this case, direct experimental verification is either difficult, expensive, or even materially impossible.
Nowadays with powerful computers, it is usually possible to simulate the experiment. In general, the agreement with experience is not total. For example, meteorology forecasts are not 100% reliable.
Consequently, there is nothing to prove that a mathematical model completely reflects reality. A fortiori, when two models are in agreement between them, that is "encouraging" but it is not an absolute proof. For this reason, the concept of validation becomes subjective. The same model can be validated by one researcher and rejected by another depending on whether the model is "judged satisfactory or not" in relation to a certain number of invoked arguments (hypotheses, approximations, etc ....).
I think it is difficult because the two have different conditions.
Hello Sir.
Actually we perform validation for a numerical model to say that this model is similar to the experimental one. Whenever the question comes to validation, the first thing is that both the numerical model and the experimental model should have the same geometry and properties. As you said about the PHE, you can validate the model with the experimental data generated by the same type. You can play with the model after validation, not before the validation. If they differ in shape and geometry, their heat transfer rate will differ, and your numerical investigation will become of no use.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Plating