Effective control of LNG regasification plant using multivariable model predictive control

Abstract

Multivariable Model Predictive Control (MMPC) is used to control temperature and pressure at the LNG regasification plant to overcome the problem of interaction between variables and reduce the number of controllers. There are four controlled variables (CV) and four manipulated variables (manipulated variables, MV). The controlled variables are the pressure on the LNG storage tank, vaporizer output pressure, vaporizer output temperature, and gas temperature towards the pipeline. The manipulated variable, which are respectively paired with the CV, are the top product flow rate of the tank, pipeline gas flow rate, incoming sea water flow rate, and duty heater. Identification of the FOPDT empirical model (First Order Plus Dead-Time) is implemented on the four pairs of CVs and MVs to describe interactions between variables. The FOPDT obtained is used as a controller in MMPC and to determine the performance of MMPC tuning parameters, namely P (prediction horizon), M (control horizon), T (sampling time). Control performance is measured using the ISE (Integral Square Error) method. As a result, the MMPC parameters (P, M, T) for optimum LNG regasification condition, respectively: 330, 1, 1. The ISE results of MMPC controller in set point tracking: 2.12×10−4, 23.834, 0.763, 0.085, with improvement of control performance respectively by 31262%, 17%, 175%, 757% compared to MPC controller performance.