I have performed a certain set of nonlinear dynamic analyses on an RC (Reinforced Concrete) frame that I designed on SAP2000 using the Indian Standard Codes. Since the numerical model is not based on any existing building, I fear that someone might ask me how I would validate my model, which on their part, is justified. What could I possibly say as a satisfactory response?
For some reason, many people are often reluctant to share their design details, so the question regarding validation seems a bit tricky.
Thanks in advance.
Hello Rohit Sinha,
to answer your question, let's start with a more general picture.
Why do we need to validate the FE or any other type of model and how can we do this?
The answer to the first part of the question is that validation is necessary because models are simplifications of real-world systems, and there may be limitations and assumptions made during the modelling process that can affect the accuracy of the results. So far, so good.
Now to the second part of the question: HOW to validate the model? And here there are different ways of doing this:
1) Compare the model's predictions with experimental measurements of the same system under the same conditions. This can involve comparing the model's predicted displacements, stresses, strains, or other quantities to the corresponding experimental measurements. But as I understand it, in your case it is impossible to do the validation experiments. Then we come to the second option,
2) Compare its predictions with those of other validated models. This can involve comparing the results of different FE models that have been developed for the same system or comparing FE models to analytical solutions or empirical data. It means that you need to do some literature research where the similar models are discussed and maybe some validations are done. So you can try to compare your results with the results in the other articles.
Aditionaly I would recommend you to discuss the assumptions of your model. You can admit that this is only a model, and not a real experiment, but the material properties, boundary conditions and applied forces correspond to real materials and processes (if so), and it means that the models results are reliable to a certain extent.
I hope these advices will help you.
Sincerely,
Pavlo
Thank you Pavlo Pavliuchenko sir for this insightful answer. This does clarify a lot of things. Of course, covid induced restrictions did play a huge role in limiting us from going out to search for experimental data or real-world data that we could have used for our FE model. However, we can surely find some literature to validate the results. Thanks
Validation is a critical aspect of any numerical model. In the case of an RC frame designed on SAP2000 using the Indian Standard Codes, there are a few ways to validate the model. Firstly, you can compare the results of your model with experimental data from similar RC frames that have been subjected to similar loading and boundary conditions. This allows you to check if your model is providing reasonable results. Secondly, you can compare the results of your model with analytical solutions based on well-established methods such as finite element analysis, finite difference analysis, or other methods. These analytical solutions are based on theoretical principles and are often used to validate numerical models. Finally, you can perform sensitivity analyses to investigate the effects of various parameters such as mesh size, time step, damping, and material properties on the results of your model. This will help you to identify the most influential parameters and ensure that your model is providing accurate results. In summary, the validation of your numerical model can be achieved through a combination of experimental data, analytical solutions, and sensitivity analyses.
- Badges
- Science topic
- Similar topics
- Analytical Chemistry
- Column