High-cell density cultures (HCDC), prone to metabolic overflow, are typically operated in fed-batch mode to maximize productivity. We developed a simulation-based procedure for assessing advanced control strategies for HCDC, aiming to avoid metabolic overflow. We tested a heuristic controller, a nonlinear model predictive controller (NMPC), and two adaptive controllers under realistic conditions. As a case study, we considered the growth of a carotenoid-producing Saccharomyces cerevisiae strain developed in our laboratory. Our control simulations were consistent with experimental results reported in the literature with equivalent control strategies. All tested controllers could cope well with measurement noise and model parameter variations, reaching low ethanol concentrations ($\le$ 1 g/L), albeit with substantially different biomass productivities. The NMPC algorithm yielded the best performance even under parametric uncertainty, achieving high biomass concentrations (200 gDCW/L) and cell productivities (5 gDCW/L/h). The presented systematic procedure helps in the evaluation of novel control strategies before their implementation in practice.

高细胞密度培养 (HCDC) 容易发生代谢溢出，通常以分批补料模式运行，以最大限度地提高生产力。我们开发了一种基于模拟的程序，用于评估 HCDC 的高级控制策略，旨在避免代谢溢出。我们在现实条件下测试了一个启发式控制器、一个非线性模型预测控制器 (NMPC) 和两个自适应控制器。作为一个案例研究，我们考虑了我们实验室开发的一种产生类胡萝卜素的酿酒酵母菌株的生长。我们的控制模拟与具有等效控制策略的文献中报告的实验结果一致。所有经过测试的控制器都能很好地应对测量噪声和模型参数变化，达到低乙醇浓度（$\le$1 g/L)，尽管具有显着不同的生物质生产力。即使在参数不确定的情况下，NMPC 算法也能产生最佳性能，实现高生物量浓度 (200 gDCW/L) 和细胞生产力 (5 gDCW/L/h)。所提出的系统程序有助于在实际实施之前评估新的控制策略。