Surrogate based optimization of a process of polycrystalline silicon production

Abstract

A hybrid polycrystalline silicon production route is optimized following a two-step procedure. First, surrogate models for the main units are developed following different techniques which depend on the available information. Secondly, the optimization of the entire process flowsheet allows determining the optimal tradeoff between yield and energy consumption. A base production capacity of 2000 t/y of polycrystalline silicon is considered, with an equipment cost of 9.97 M$. Three scenarios are evaluated: maximum silicon production, minimum operating costs and maximum total profit. The maximization of the total profit is the most promising scenario, obtaining a selling price of 8.93 $/kgPoly, below the commercial price, 10 $/kgPoly. The revenue obtained is 10 M$/y, with an operating cost of 6.48 M$/y. Furthermore, a plant scale-up study was performed. If the production capacity is increased by a factor of 10, it results in a reduction of 1.03 $/kgSiPoly.