Enhanced monitoring of linear profiles in dynamic processes using advanced memory-type control charts

In chemical processes, it is customary to deal with the chemical reactors that function with a continuous flow of reactants and products while preserving a constant composition inside the reactor. It involves addressing two challenging issues including state estimation and the attenuation of external disturbances. A suitable control method is quite challenging to design because of the tremendous complexity of such processes. It is essential to monitor even the minor alterations in these processes to guarantee satisfactory performance. Statistical process control (SPC) provides such monitoring tools in the form of control charts that help to maintain the stability of these processes. The process monitoring in SPC often involves a profile function that influences the performance of the process or product, which can be described by a model. This study presents new methods based on double exponentially weighted moving average statistics to enhance the monitoring of the linear profiles. The proposed design structure covers three memory charts to monitor the intercept, slope, and error’s variance. The effectiveness of the proposals is evaluated using various run length-based performance measures such as average run length and conditional expect delay under zero and steady states. A comparison is made between the proposed and competing methods, and their superiority is demonstrated based on their ability to identify shifts in various process parameters. The comparative examination demonstrated that the proposed structure outperforms the competing charts to detect small shifts in the parameters of the profile model. For practical considerations, the implementation of the study proposal is shown for a continuous stirred-tank reactor (CSTR) process that is widely used by the chemical industry, as well as other manufacturing industries.