However, global optimization methods generally require up to tens of thousands of model runs to find the global optimal solution. This may place severe computational constraint on solving such an optimization problem, if the underlying model requires a large amount of CPU time to run. One approach to reduce the computational burden is to approximate and replace the expensive simulation model with a cheaper-to-run surrogate model. There are two broad families of surrogates: (1) response surface surrogates, which are statistical or empirical data-driven models emulating the responses of a high-fidelity simulation model; and (2) lower-fidelity physical based surrogates, which are simplified models of the original system