Validation of Class of Non-Linear Controllers on Lab Scale Batch Reactor: Sliding Mode Controllers

This article aims to implement the sliding mode controllers (SMCs), which is a non-linear controller with high computational intensity. Also, the implementation of various non-linear controllers such as model predictive controller (MPC), Non-linear model predictive controllers (NMPC) involves online optimization with constraints. Hence the validation on physical system has limitations on computational capacity for online data. In this article, authors have addressed the overcoming of high intense computations using Jetson developer kit with signal conditioning and analog to digital converters (ADC) to connect the physical sensor and actuator signals. Case study of highly nonlinear batch reactors is considered. Due to its exothermic reaction with chemical feed along with its non-steady state. The uncertainty in this batch reactor experimental study is considered as the change in input cold water, which is circulated in the reactor jacket to carry away the additional heat. SMCs are more robust and sensitive to additional uncertainty added into the system dynamics. In this paper authors considered the nonlinear model equations developed for the experimental batch reactor by Prajwal et al. [1, 2]. The optimal temperature profile is developed for acrylamide polymerization reaction and used as a reference trajectory for the closed loop simulation and validation. The coolant flow rate is the manipulated variable circulated via circular coil and the reactor temperature is the process variable. In this paper authors have presented four types of SMC to arrest the reactor temperature oscillations with the reference trajectory. The experimental results depict that the SMC with PI sliding surface gives minimum control signal and reactor temperature tracks the trajectory over a period of time than other types of SMC. The major bottleneck in implementing advanced controllers with constrains such as NMPC is the computation time with optimizer. Attempt of implementing advanced computations techniques are in progress using Jetson board.