Abstract

A non-segregated kinetic model is proposed to describe a fermentation process of agro-industrial residues derived via cocoa (mucilage juice) by Pichia kudriavzevii. The novel proposed hybrid model is based on a multiple coupling reaction mechanisms (structured) to describe the kinetics of substrate consumption, biomass, carbon dioxide, and ethanol, coupled to an unstructured model for the activity enzyme. The parameters of the kinetic model are estimated by non-linear least-squares curve fitting using the Marquardt-Levenberg algorithm. In addition, numerical simulations were compared with the experimental data via residual graphs. The effectiveness of the model was statistically evaluated using dimensionless efficiency coefficients under different initial conditions. A global sensitivity analysis was applied (Fisher's information matrix). The experimental results of the batch reactor showed a maximum ethanol concentration of 29 g/L, with a yield of 0.48 g-ethanol/g-glucose and a productivity of 0.30 g/L h. The method determined that the cell formation coefficient and the specific substrate consumption rate (${\theta }_1$ and ${\theta }_2$) directly influence most of the states of our system. The proposed scheme is particularly suitable to assist in the rational design of cell factory properties or fermentation processes because it can represent the complex biochemistry in more detail and under different initial experimental conditions; the above reveals that the generated model is robust and can be considered for control and optimization purposes.

摘要

提出了一种非分离动力学模型来描述Pichia kudriavzevii对可可（粘液）衍生的农业工业残留物的发酵过程。新提出的混合模型基于多重耦合反应机制（结构化）来描述底物消耗、生物质、二氧化碳和乙醇的动力学，并耦合到活性酶的非结构化模型。使用 Marquardt-Levenberg 算法通过非线性最小二乘曲线拟合估计动力学模型的参数。此外，通过将数值模拟与实验数据进行比较残差图。在不同的初始条件下，使用无量纲效率系数对模型的有效性进行统计评估。应用了全局敏感性分析（Fisher 信息矩阵）。间歇式反应器的实验结果表明，最大乙醇浓度为 29 g/L，产量为 0.48 g-乙醇/g-葡萄糖，生产率为 0.30 g/L·h。该方法测定细胞形成系数和比底物消耗率（${\theta }_{\mathrm{1个}}$和${\theta }_{\mathrm{2个}}$) 直接影响我们系统的大部分状态。所提出的方案特别适合协助细胞工厂属性或发酵过程的合理设计，因为它可以更详细地表示复杂的生物化学，并且在不同的初始实验条件下；以上表明生成的模型是稳健的，可以考虑用于控制和优化目的。