| Abstrakt: |
Transfer of laboratory-scale experiments to production-scale ethanol fermentation is time-consuming and involves expensive prototype systems from complex experimental designs that determine optimal operating conditions for minimal substrate and product inhibitions. The study developed and validated a Simulink-based model for optimal pH and temperature control using fuzzy logic and PID controllers respectively and taking advantage of 2D and 1D substrate and product inhibition models from which suitable ethanol fermentation reaction rates models were selected. Temperature and pH levels and substrate, product, and biomass concentrations were measured. Selected inhibition models were linear-product, linear substrate-sudden stop product, and linear substrate for cassava, maize, and sorghum, respectively. Fuzzy logic controller ascertained optimal flow rate of acid and base as 0.000196 ml/s and 0.000204 ml/s, respectively, and pH error and rate of pH error as 0.00334 and 0.00368, respectively. F-test two-sample for variances showed no significant difference between model and experimental curves (cassava: F critical = 0.9704, F calculated = 0.1905; maize: F critical 0.9704, F calculated = 0.2149; sorghum: F critical = 0.9704, F calculated = 0.2488). PID logic controller showed model curves and experimental curves with good fit. F-test two-sample for variances showed no significant difference between model and experimental curves (cassava: F critical = 0.9704, F calculated = 0.1288; maize: F critical = 0.9704, F calculated = 0.2083; sorghum: F critical = 0.9704, F calculated = 0.2016). The study provided an improved approach as solution for optimal pH and temperature conditions in order to mitigate substrate and product inhibitions during ethanol fermentation. It illustrated that the application of artificial intelligence-based controllers provides satisfactory outcomes that are desirable for implementation in the industrial space. Highlights: ✓ Inhibition models were selected for three feed stocks using the model fitness coefficient and applied to ethanol fermentation batch dynamic model. ✓ Selected inhibition models included the following: linear-product, linear substrate-sudden stop product, and linear substrate for cassava, maize, and sorghum, respectively. ✓ Fuzzy logic controller determined optimal pH profile, flow rates of acid and base, pH error, and rate of pH error. ✓ There was no significant difference between model and experimental profiles (cassava: F critical = 0.9704, F calculated = 0.1905; maize: F critical 0.9704, F calculated = 0.2149; sorghum: F critical = 0.9704, F calculated = 0.2488). ✓ PID logic controller determined optimal temperature profile. ✓ There was no significant difference between model and experimental profiles (cassava: F critical = 0.9704, F calculated = 0.1288; maize: F critical = 0.9704, F calculated = 0.2083; sorghum: F critical = 0.9704, F calculated = 0.2016). [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink