Conference abstract – PCCC8: Demonstration of economic nonlinear model predictive control of CO2 capture plant with CESAR1

Abstract

A nonlinear model predictive control (NMPC) strategy is successfully demonstrated for the Tiller CO2Lab (SINTEF AS) aminebased post-combustion CO2 capture pilot using the CESAR1 solvent. The dynamic pilot plant model is validated against experimental results with good agreement. Building on previous experience with testing this type of controller, a particular focus has been to combine day-to-day economic optimization with supervisory control of the plant performance. Such control structure consists of two hierarchical layers, an underlying NMPC layer for supervisory control subject to varying operational process conditions, and a dynamic real-time optimizer (DRTO) on top of the NMPC whose role is to account for slower, day-to-day variability. Specifically, reboiler energy input and solvent flow rate are manipulated to control the CO2 capture rate in the absorber and the specific reboiler duty (SRD), while the economic optimization is connected to varying price of electricity and CO2 tax. The total operational cost is minimized over the 24-hour horizon to maintain an average CO2 capture rate of 92% or higher. Since this kind of automatic control under varying operating regimes is able to optimize multiple variables simultaneously, it enables considerable cost savings compared to manual control. This flexibility is especially significant as energy variability (both in terms of cost and availability) is expected to increase as a result of the introduction of renewables to the energy mix. Not only can the energy usage and costs related to CO2 tax be reduced, but with a control structure as demonstrated here, the plant operation will also be less labor intensive.

Keywords: Post combustion carbon capture; amine based CO2 absorption; CESAR1 Solvent; nonlinear model predictive control; optimal control; OPEX reduction; flexible operation

Authors: Adriaen Verheyleweghen (Cybernetica AS), Fredrik Gjertsen (Cybernetica AS), Thor Mejdell (SINTEF Industry), Hanne M. Kvamsdal (SINTEF Industry).