A parallel-flow regenerative kiln is an efficient calcination process for the production of quicklime. It is a complex process where two connecting shafts work in tandem with alternating burning and regeneration cycles that improve the calcination efficiency in contrast with single shaft and rotary technologies. For the control and optimization of this process, it is necessary to have accurate models of the kiln dynamics and different interacting components. This paper presents a methodology to select the appropriate signals to produce data-driven models of the kiln as a linear interacting system.

This paper is dedicated to the experimental validation of a cascade control strategy for simultaneously regulating the glucose and biomass levels in continuously perfused HEK-293 cell cultures. The inner loop consists of a partial feedback linearization, which requires the estimation of the biomass specific growth rate and glucose uptake rate. This latter task is achieved by sliding mode observers, which do not require a priori process knowledge in the form of a process model. The linearized process is then regulated by the outer loop, including two classical PI controllers with autotuning. The four manipulated variables are a feed flow rate with low glucose concentration, another feed flow with a higher glucose content, a bleed flow, and a perfusion stream. The experimental results demonstrate the ability of the control strategy to reach and regulate the prescribed setpoints. The main advantage of the strategy is that it can be applied in a plug and play manner and shows satisfactory robustness. To the best of our knowledge, this is the first time that such a multivariable control strategy, together with sliding mode observers, is applied at the lab scale to an industrial process in the pharmaceutical sector.

Measured external fluxes impose constraints to under-determined metabolic networks that narrow the internal flux intervals obtained using Flux Variability Analysis. Nevertheless, these constraints often lead to systems that do not admit a feasible solution. Measurement noise and data smoothing are among the sources of uncertainties that can cause system infeasibility. These constraints are classically released using interval representation of fluxes. This study investigates the use of Adaptive Flux Variability Analysis (AFVA), which allows determining a minimal coefficient of variation of the external fluxes along the time course of the experiment. Especially, AFVA is applied to a medium-size metabolic network and a rich dataset relative to HEK-293 cells cultured in batch, encompassing all 20 amino acids and less commonly measured metabolites, such as urea and pyruvate. AFVA appears as an effective tool for metabolic flux analysis. The impact of data-smoothing and the information provided by the cell growth are thoroughly analyzed.

Underdetermined metabolic networks are usually investigated using Flux Variability Analysis (FVA) under the pseudo-steady state assumption of the internal metabolites. When a dynamic overview of the flux map is sought, the time variation of the cell uptake and excretion rates is deduced from extracellular dynamic mass balances and smoothing of the measurement data of the time evolution of the extracellular concentrations. Nevertheless, the resulting system of equations does not always admit a feasible solution under these constraints. Indeed, measurement data is affected by noise, whose processing (smoothing, etc) always entails some subjective user choices, and the network itself might not be perfectly suited to explain the several culture phases. To alleviate these adverse effects, the constraints can be relaxed by introducing coefficients of variation of the external fluxes, and by considering an interval representation of the fluxes. This work presents a systematic method to determine these coefficients of variation, along the time course of the culture, leading to an adaptive scheme where the coefficients are set to the tightest bounds. The methodology is applied to experimental data of cultures of hybridoma in batch and perfusion modes and compared to previously published results.

This study aims at providing a methodology based on both data- and knowledge-driven approaches to build dynamic macroscopic models of cell cultures. This methodology proceeds in three steps. A principal component analysis is first applied in order to determine the minimum number of macroreactions necessary to faithfully describe the available data. These reactions are then selected among the elementary flux modes associated with a chosen metabolic network through the definition of an original linear programming problem. Kinetic laws are finally identified so as to reproduce the measurement data. The proposed methodology is illustrated using four different perfusion cultures of hybridoma cells and demonstrates a good capacity to select macroreactions capable of reproducing well the complex experimental data, even with the use of simple kinetic laws and without re-identifying the stoichiometry.

In this study, a dynamic model of a Vero cell culture‐based dengue vaccine production process is developed. The approach consists in describing the process dynamics as functions of the whole living (uninfected and infected) biomass whereas previous works are based on population balance approaches. Based on the assumption that infected biomass evolves faster than other variable, the model can be simplified using a slow‐fast approximation. The structural identifiability of the model is analysed using differential algebra as implemented in the software DAISY. The model parameters are inferred from experimental datasets collected from an actual vaccine production process and the model predictive capability is confirmed both in direct and cross‐validation. The model prediction shows the impact of the metabolism on virus yield and confirms observations reported in previous studies. Multi‐modality and sensitivity analysis complement the parameter estimation, and allow to obtain confidence intervals on both parameters and state estimates. Finally, the model is used to compute the maximum infectious virus yield that can be obtained for different combinations of multiplicity of infection (MOI) and time of infection (TOI). © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 35: e2687, 2019

In this work, a dynamic model of infection and virus amplification in vero cell cultures targeting vaccine production is proposed. In contrast with previous works, the model describes the process dynamics as functions of the whole living (uninfected and infected) biomass. The dynamic model is based on a slow-fast approximation where the infected biomass is considered as evolving faster than other variables. The resulting model contains unknown parameters that are inferred from datasets collected from an actual vaccine production process. Parameter identification is complemented by a sensitivity analysis and the determination of confidence intervals for the parameters and predicted trajectories. Results are in general in good agreement with the experimental data.