Investigating the optimal preference of multi-substrate for xylitol production using unstructured mathematical modelling

Abstract

Abstract An unstructured kinetic model was developed to describe xylitol fermentation by Candida tropicalis using enzymatic hydrolysate derived from pineapple core waste as a lignocellulosic feedstock. The model was shown to satisfy the mathematical conditions of positivity and boundedness, supporting its numerical consistency. Predictive capability was evaluated through leave one out cross validation on the available batch fermentation dataset. The model showed good predictive performance for glucose, xylose, and xylitol concentrations ( R 2 = 0.93–0.96), while biomass prediction was less accurate due to the growth to death transition observed during the later stage of fermentation. Parameter identifiability analysis indicated that several kinetic parameters, particularly those related to xylitol production, could not be uniquely estimated from the single batch dataset. Global sensitivity analysis revealed that biomass and substrate dynamics were mainly governed by the cell death rate constant ( k d ) and growth parameters, whereas xylitol production was strongly influenced by non-growth associated pathways ( β P , m 3 ) and growth-related xylose consumption ( α L ). Glucose pathway parameters exerted negligible influence. Consistent findings from identifiability and sensitivity analyses suggested that model predictions depended more strongly on parameter combinations than on individual parameter values. An optimal control strategy derived from the model increased simulated xylitol and biomass accumulation by limiting excessive glucose utilization, consistent with reduced carbon catabolite repression. These results highlight the value of integrating identifiability, sensitivity, and control analyses within a kinetic modeling framework and suggest the potential of pineapple core waste as a sustainable substrate for xylitol bioproduction.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint