I would like to perform a sensitivity analysis of a CFD solver. There are 8 input variables, for each of them there are 2-3 prescribed numerical values.
To evaluate one set of parameters three costly simulations (each running for 20 hours on 800 cpu cores). Budget for these simulations is limited and due to the queuing system of the HPC, it would take a long time to get the results.
I'm aware of latin hypercube hierarchical refinement methods that allows starting the sensitivity analysis with smaller budget and subsequently incorporating newer results when they're available.
But those methods works with continuous variables. Is there a method for categorical and ranked/ordinal variables?
Hello Vojtech,
can you please elaborate a little more your question? In particular, are the 8 input variables discrete (e.g. number of flaps of a wing, number of injectors in a combustion chambers), or they are continuous but you have 2-3 prescribed values (e.g. Mach number, pressure, altitude).
AG
Hello Andrea,
all but one variable has prescribed values from continuous intervals. The one remaining variable is purely categorical.
Vojtech
Hello Vojtech,
did you attempt to use orthogonal array (OA) designs?
With OA you can specify the number of levels of each input (input values), and the resulting design is usually a good starting point. If then you would like to increase the coverage of your design because you have more time budget, you can easily do it based on where you would like more information.
If you have limited simulation budget, and most of input variables are discrete, I would invest some time on OA.
Hello Vojtech,
a good reference for most of classical and modern design of experiment techniques is:
A. Atkinson, A. Donev, and R. Tobias,Optimum experimental designs, with SAS.Oxford University Press, 2007, vol. 34
Another good book is:
T. J. Santner, B. J. Williams, and W. I. Notz,The design and analysis of computerexperiments. Springer Science & Business Media, 2013
- Badges
- Science topic
- Similar topics
- Engineering
- Control Theory
- Control Systems
- Control Systems Engineering