The use of metamodels allows the optimization of steady-state flowsheet simulations to proceed while requiring only a small number of solutions to be obtained from the simulation. This paper presents the techniques needed to optimize the metamodel forms described by Palmer and Realff. Through a case study, we exemplify characteristics of the solution process and demonstrate the strength of the metamodeling approach. The operating cost of an ammonia synthesis plant is optimized over six input variables using a total of only 32 solutions of the plant simulation as data for the metamodels.