Value chain planning optimization: A data driven digital twin approach

The long term production planning for a large chemical production site, where 10+ different chemical plants share raw materials, infrastructure (e.g., tank farm, filling stations) and utilities (e.g. steam, electricity, technical gasses) might prove to be a challenging task. This paper introduces a data driven approach to build a digital twin of a chemical production site to aid the relevant decision makers in defining and evaluating the economic impact of a long term (i.e. several months ahead) production planning. Each chemical plant and energy production unit on site is represented by simple regression models relating the consumption of raw materials and utilities to its products. The resulting system of algebraic equations has been inserted in an optimization environment with the objective of maximizing the profit. In the optimization, also the electricity and steam generation were introduced to obtain a global energy balance of the production site. This combination resulted in a multi period Mixed-Integer Linear Programming (MILP) problem. The effect of electricity price and external temperature on the optimization results are also investigated.