Optimizing Carboxymethyl Cellulose (CMC) Production: A Metaheuristic Algorithm Approach

Due to the characteristics, including surface properties, mechanical strength, tunable hydrophilicity, and others, Carboxymethyl cellulose or CMC is widely used in various advanced applications, namely food, paper, textile, pharmaceutical industries, biomedical engineering, wastewater treatment, energy production, and others. Consequently, producing a higher amount of CMC remains a significant research challenge since it involves several key factors, namely Raw Material Selection, Reaction Conditions, Alkali Concentration, Degree of Substitution, and others. Continuous monitoring and adjustment of these key factors can enhance yield and quality. Typically, this process is carried out manually in the lab, which can be both time-consuming and expensive. However, advanced techniques such as computer-based process analytics and modeling, as discussed in this paper, can facilitate optimization. It is well-known fact that the parameters involved in CMC generation often exhibit non-linear behavior, and therefore, meta-heuristic optimization algorithms combined with machine learning techniques can offer improved performance. Therefore, this paper employs several meta-heuristic optimization algorithms in conjunction with the Random Forest Regressor or RFR to identify the optimal parameters and its optimal values for enhancing CMC production. This proposed technique is applied on laboratory experimental data. Boxplot analysis reveals that optimal CMC production averages are achieved with chemical amounts ranging from 0.5 to 2.0 for the Sine Cosine Algorithm (SCA), yielding an average of 12, and from 0.3 to 1.0 for the Particle Swarm Optimization (PSO), yielding an average of 9. The results acquired from the experiments demonstrate that the metaheuristic algorithms can find out the optimal parameters and their values for increasing CMC generation.