Black-box modelling, bi-objective optimization and ASPEN batch simulation of phenolic compound extraction from root.

root (NLR) extract is one of phytochemicals used to treat various ailments in most of developing countries. This investigation focuses on modelling, optimization and computer-aided simulation of phenolic solid-liquid extraction from NLR. The extraction experiments were conducted at extraction temperature (ET: 33.79-76.21 °C), process time (PT: 2.79-4.21 h) and solid-liquid ratio (SLC: 0.007929-0.018355 g/ml). Regression models (RM) were developed, using Response Surface Methodology (RSM) in Design Expert software, for predicting and optimizing total phenolic content (TPC) and total flavonoid content (TFC) and also compared with adaptive neuro-fuzzy inference system (ANFIS) modelling in Matlab environment. Aspen Batch Process Developer (ABPD) V10 was used to simulate phenolic extract production and perform material balance of the process. Both Coefficients of determination (R) of RSM (TFC: 0.9996, TPC: 0.9932) and ANFIS models (TFC: 0.99998, TPC: 0.9982) were compared and predicted satisfactorily. Optimization results show: ET (2.79 h), PT (38.8 °C), SLC (0.0198 g/ml), TFC (25.92 25.92 μg RE/g) and TPC (8.47 mg GAE/g). The phenolic extraction base case simulation results gave batch throughput, annual throughput, number of batches per year 0.0089 g/batch, 0.139 g/year and 1019 batches, respectively. The ABPD predicted TPC and experimental TPC results were compared and gave mean relative deviation error of 3.75%. Thus, ABPD simulation model is reasonably reliable for the scale-up design engineering of the phenolic extract production from NLR.