Modelling and Optimization of Primary Steam Reformer System Case Study: the Primary Reformer PT Petrokimia Gresik Indonesia

Abstrak

Steam reforming of hydrocarbons has been in use as the principal process for the generation of hydrogen and synthesis gas needed in the Ammonia production in petrochemical industries. Optimal operation of existing steam reformers is crucial in view of the high energy consumption and large value addition involved in the process. The economic objective of the process is determined by the cost of gas (Methane), the cost of steam and additional fuel. An optimum steam to gas ratio is expected from an optimum process control.  This can be applied on ratio control parameter for natural gas feed. In addition, steam Carbon Ratio is used to decrease coke formation in catalyst reformer. This paper presents the model identification and optimization of reforming control system of an industrial Primary Reformer at PT. Petrokimia Gresik (One of the fertilizer petrochemical industry in Indonesia). The reformer model has been approximated in the form of Takagi-Sugeno-Kang fuzzy inference system, with architecture in neural-network model. ANFIS (Adaptive Neuro-Fuzzy Inference System) has been utilized to determine NARX or ARX parameters model describing the dynamic of Industrial operational data have been used for training and validating the model. The optimization problem has been addressed through the utilization of Constrained Nonlinear Programming. The aim is to find the optimal process and ratio controller parameters to achieve the maximum Hydrogen formation. For a maximum fixed production rate of hydrogen produced by the unit, minimization of methane feed rate is chosen as the objective function to meet processing requirements.

Keywords: Hydrocarbons, Model identification, ANFIS