I am going to conduct a numerical simulation based on CFD method for solving a fluid mechanics problem. Based on the significant factors contributing to the study, the number of simulation is too much!!!! Previously, I had an experience to reduce the number of tests in an experimental study through the Taguchi algorithm. I am wondering that, whether or not I can use the similar algorithm in a numerical (CFD) work to reduce the number of simulations?
I have done this method previously in FE method and Multi body dynamic analysis. The results will be amazing and you can find many significant outcomes from your simulation. However, it is recommended to perform some confirmation test to validate your ANOVA table and empirical model.
Tx Dr. Farokhinejad
What's that supposed you mean from the confirmation tests? extra simulation to confirm the validity of primary simulations' results?
You will confirm the simulation results with some test like T and Z test which is used in DOE. I agree with the answer of Ali Faroki
Dear roozbeh,
In numerical simulation test you dont have replication and center points. Therfore, after carrying out the ANOVA analysis you will have an emperical model. You should chose a number in range of variables but not exact number of variable ( a random number between upper and lower bound).
Then you calculate the corresponding response. After that you run your simulation based on the chosen numbers. If the DoE results matches with numerical results it means that your simulation confirm your statistical model. This method is very beneficial when you chose non linear model such as RSM , Latin hyper cube or Taguchi models.
- Badges
- Science topic