Simulation of pH level distribution inside a coiled flow inverter for continuous low pH viral inactivation

The continuous production of monoclonal antibodies (mAbs) with the help of disposable equipment poses one of the future major changes in the pharmaceutical industry. Consequently, continuous viral clearance needs to be developed as well. The coiled flow inverter (CFI) was successfully implemented in the continuous downstream as a residence time module for low pH viral inactivation. As the elution profile of the upstream continuously operated protein A chromatography results in fluctuating pH values, the pH level distribution inside the CFI is highly relevant. This study presents a detailed investigation of pH level distribution inside the CFI at varying inlet conditions with the help of computational fluid dynamics simulation. The simulation model was validated first with the help of experimental data. Afterwards, the model was used for further investigations. It was determined that with a pH sine curve as input, the duration until steady state at the outlet requires two times the minimum residence time of the apparatus. Moreover, it could be observed that the CFI itself offers a progressive dampening effect on the pH level distribution. Afterwards, different forms of the sine curve representing different operation modes of the continuous protein A chromatograph were tested to evaluate this dampening capability. It became clear that the switch time has the highest influence on the resulting pH of the outlet stream and should be considered for process development. Finally, the radial pH profiles at different positions inside the CFI were determined. This once again revealed the high radial mixing capability of the CFI and its influence on the resulting product stream.     

Virus study for continuous low pH viral inactivation inside a coiled flow inverter

Continuous processing for the production of monoclonal antibodies (mAb) gains more and more importance. Several solutions exist for all the necessary production steps, leading to the possibility to build fully continuous processes. Low pH viral inactivation is a part of the standard platform process for mAb production. Consequently, Klutz et al. introduced the coiled flow inverter (CFI) as a tool for continuous low pH viral inactivation. Besides theoretical calculations of viral reduction, no viral clearance study has been presented so far. In addition, the validation of continuous viral clearance is often neglected in the already existing studies for continuous processing. This study shows in detail the development and execution of a virus study for continuous low pH viral inactivation inside a CFI. The concept presented is also valid for adaptation to other continuous viral clearance steps. The development of this concept includes the technical rationale for an experimental setup, a valid spiking procedure, and finally a sampling method. The experimental results shown represent a viral study using xenotropic murine leukemia virus as a model virus. Two different protein A (ProtA) chromatography setups with varying pH levels were tested. In addition, one of these setups was tested against a batch experiment utilizing the same process material. The results show that sufficient low pH viral inactivation (decadic logarithm reduction value >4) was achieved in all experiments. Complete viral inactivation took place within the first 14.5 min for both continuous studies and the batch study, hence showing similar results. This study therefore represents a successful virus study concept and experiment for a continuous viral inactivation step. Moreover, it was shown that the transfer from batch results to the continuous process is possible. This is accomplished by the narrow residence time distribution of the CFI, showing how close the setup approaches the ideal plug flow and with that batch operation.     

Fluid flow in a spiral device used for irradiation of biological fluids

The manufacture of plasma‐derived therapeutics includes dedicated viral inactivation steps to minimize the risk of infection. Traditional viral inactivation methods are effective for the removal and inactivation of enveloped viruses, but less effective against small nonenveloped viruses. UV‐C irradiation has been demonstrated to be an effective means of inactivating such viruses. The UVivatec lab system consists of a spiral tube around an UV‐C irradiation source. Flow of a solution through the chamber generates and ensures controlled mixing and uniform exposure to irradiation. A detailed assessment of the effect of flow rate, alternate cross sectional design and scale up of the irradiation chamber on Dean vortices was performed using the smoothed particle hydrodynamics method. The aim was to provide a basis for setting flow rate limits and using a laboratory scale apparatus to model viral inactivation in larger manufacturing scale equipment. The effect of flow rate related changes on the fluence rate was also investigated through chemical actinometry studies. The data were consistent with the simulations indicating that Dean vortices were present at low flow rates, but dissipated at higher flow rates through the spiral chamber. Importantly, this work also allowed a correlation between the small system and large scale system to be established. This will greatly facilitate process development and viral validation studies. © 2013 American Institute of Chemical Engineers Biotechnol. Prog., 29: 359–367, 2013     

Validation of CFD predictions of flow in a 3D alveolated bend with experimental data

Verifying numerical predictions with experimental data is an important aspect of any modeling studies. In the case of the lung, the absence of direct in vivo flow measurements makes such verification almost impossible. We performed computational fluid dynamics (CFD) simulations in a 3D scaled-up model of an alveolated bend with rigid walls that incorporated essential geometrical characteristics of human alveolar structures and compared numerical predictions with experimental flow measurements made in the same model by particle image velocimetry (PIV). Flow in both models was representative of acinar flow during normal breathing (0.82ml/s). The experimental model was built in silicone and silicone oil was used as the carrier fluid. Flow measurements were obtained by an ensemble averaging procedure. CFD simulation was performed with STAR-CCM+ (CD-Adapco) using a polyhedral unstructured mesh. Velocity profiles in the central duct were parabolic and no bulk convection existed between the central duct and the alveoli. Velocities inside the alveoli were approximately 2 orders of magnitude smaller than the mean velocity in the central duct. CFD data agreed well with those obtained by PIV. In the central duct, data agreed within 1%. The maximum simulated velocity along the centerline of the model was 0.5% larger than measured experimentally. In the alveolar cavities, data agreed within 15% on average. This suggests that CFD techniques can satisfactorily predict acinar-type flow. Such a validation ensure a great degree of confidence in the accuracy of predictions made in more complex models of the alveolar region of the lung using similar CFD techniques.     

Experimental and numerical study of heterogeneous pressure-temperature-induced lethal and sublethal injury of Lactococcus lactis in a medium scale high-pressure autoclave

The present contribution is dedicated to experimental and theoretical assessment of microbiological process heterogeneities of the high-pressure (HP) inactivation of Lactococcus lactis ssp. cremoris MG 1363. The inactivation kinetics are determined in dependence of pressure, process time, temperature and absence or presence of co-solutes in the buffer system namely 4 M sodium chloride and 1.5 M sucrose. The kinetic analysis is carried out in a 0.1-L autoclave in order to minimise thermal and convective effects. Upon these data, a deterministic inactivation model is formulated with the logistic equation. Its independent variables represent the counts of viable cells (viable but injured) and of the stress-resistant cells (viable and not injured). This model is then coupled to a thermo-fluiddynamical simulation method, high-pressure computer fluid dynamics technique (HP-CFD), which yields spatiotemporal temperature and flow fields occurring during the HP application inside any considered autoclave. Besides the thermo-fluiddynamic quantities, the coupled model predicts also the spatiotemporal distribution of both viable (VC) and stress-resistant cell counts (SRC). In order to assess the process non-uniformity of the microbial inactivation in a 3.3-L autoclave experimentally, microbial samples are placed at two distinct locations and are exposed to various process conditions. It can be shown with both, experimental and theoretical models that thermal heterogeneities induce process non-uniformities of more than one decimal power in the counts of the viable cells at the end of the treatment.     

Truly continuous low pH viral inactivation for biopharmaceutical process integration

Continuous virus inactivation (VI) has received little attention in the efforts to realize fully continuous biomanufacturing in the future. Implementation of continuous VI must assure a specific minimum incubation time, typically 60 min. To guarantee the minimum incubation time, we implemented a packed bed continuous viral inactivation reactor (CVIR) with narrow residence time distribution (RTD) for low pH incubation. We show that the RTD does not broaden significantly over a wide range of linear flow velocities—which highlights the flexibility and robustness of the design. Prolonged exposure to acidic pH has no impact on bed stability, assuring constant RTD throughout long term operation. The suitability of the packed bed CVIR for low pH inactivation is shown with two industry-standard model viruses, that is xenotropic murine leukemia virus and pseudorabies virus. Controls at neutral pH showed no system-induced VI. At low pH, significant VI is observed, even after only 15 min. Based on the low pH inactivation kinetics, the continuous process is equivalent to traditional batch operation. This study establishes a concept for continuous low pH inactivation and, together with previous reports, highlights the versatility of the packed bed reactor for continuous VI, regardless of the inactivation method.     

Validation of CFD simulations of cerebral aneurysms with implication of geometric variations

BACKGROUND: Computational fluid dynamics (CFD) simulations using medical-image-based anatomical vascular geometry are now gaining clinical relevance. This study aimed at validating the CFD methodology for studying cerebral aneurysms by using particle image velocimetry (PIV) measurements, with a focus on the effects of small geometric variations in aneurysm models on the flow dynamics obtained with CFD. METHOD OF APPROACH: An experimental phantom was fabricated out of silicone elastomer to best mimic a spherical aneurysm model. PIV measurements were obtained from the phantom and compared with the CFD results from an ideal spherical aneurysm model (S1). These measurements were also compared with CFD results, based on the geometry reconstructed from three-dimensional images of the experimental phantom. We further performed CFD analysis on two geometric variations, S2 and S3, of the phantom to investigate the effects of small geometric variations on the aneurysmal flow field. Results. We found poor agreement between the CFD results from the ideal spherical aneurysm model and the PIV measurements from the phantom, including inconsistent secondary flow patterns. The CFD results based on the actual phantom geometry, however, matched well with the PIV measurements. CFD of models S2 and S3 produced qualitatively similar flow fields to that of the phantom but quantitatively significant changes in key hemodynamic parameters such as vorticity, positive circulation, and wall shear stress. CONCLUSION: CFD simulation results can closely match experimental measurements as long as both are performed on the same model geometry. Small geometric variations on the aneurysm model can significantly alter the flow-field and key hemodynamic parameters. Since medical images are subjected to geometric uncertainties, image-based patient-specific CFD results must be carefully scrutinized before providing clinical feedback.     

A single amino acid substitution in the capsid of foot-and-mouth disease virus can increase acid resistance

Foot-and-mouth disease virus (FMDV) particles lose infectivity due to their disassembly at pH values slightly below neutrality. This acid-dependent disassembly process is required for viral RNA release inside endosomes. To study the molecular determinants of viral resistance to acid-induced disassembly, six FMDV variants with increased resistance to acid inactivation were isolated. Infection by these mutants was more sensitive to drugs that raise the endosomal pH (NH(4)Cl and concanamycin A) than was infection by the parental C-S8c1 virus, confirming that the increase in acid resistance is related to a lower pH requirement for productive uncoating. Amino acid replacement N17D at the N terminus of VP1 capsid protein was found in all six mutants. This single substitution was shown to be responsible for increased acid resistance when introduced into an infectious FMDV clone. The increased resistance of this mutant against acid-induced inactivation was shown to be due to its increased resistance against capsid dissociation into pentameric subunits. Interestingly, the N17D mutation was located close to but not at the interpentamer interfaces. The mutants described here extend the panel of FMDV variants exhibiting different pH sensitivities and illustrate the adaptive flexibility of viral quasispecies to pH variations.     

Development and validation of computational fluid dynamics models for prediction of heat transfer and thermal microenvironments of corals

We present Computational Fluid Dynamics (CFD) models of the coupled dynamics of water flow, heat transfer and irradiance in and around corals to predict temperatures experienced by corals. These models were validated against controlled laboratory experiments, under constant and transient irradiance, for hemispherical and branching corals. Our CFD models agree very well with experimental studies. A linear relationship between irradiance and coral surface warming was evident in both the simulation and experimental result agreeing with heat transfer theory. However, CFD models for the steady state simulation produced a better fit to the linear relationship than the experimental data, likely due to experimental error in the empirical measurements. The consistency of our modelling results with experimental observations demonstrates the applicability of CFD simulations, such as the models developed here, to coral bleaching studies. A study of the influence of coral skeletal porosity and skeletal bulk density on surface warming was also undertaken, demonstrating boundary layer behaviour, and interstitial flow magnitude and temperature profiles in coral cross sections. Our models compliment recent studies showing systematic changes in these parameters in some coral colonies and have utility in the prediction of coral bleaching.     

Inactivation of enzymes in fresh sake using a continuous flow system for high-pressure carbonation

The Inactivation kinetics of alpha-glucosidase, glucoamylase, alpha-amylase, and acid carboxypeptidase in fresh sake using a continuous flow system for high-pressure carbonation were investigated. In addition, the effects of ethanol and sugar concentrations on inactivation of the enzymes in high-pressure carbonated sake were investigated. Among the enzymes investigated, alpha-glucosidase was the most stable and alpha-amylase was the most labile on inactivation under carbonation. The decimal reduction times (D values) of alpha-glucosidase, glucoamylase, alpha-amylase (extrapolated from the Z value), and acid carboxypeptidase were 29, 6, 2, and 5 min respectively at 45 degrees C. These values are lower than those subjected to heat treatment. On the carbonation treatment as well as the heat treatment, ethanol accelerated the inactivation of all four enzymes, but glucose depressed the inactivation of these enzymes, except for acid carboxypeptidase. These results suggest that this continuous flow system enabled effective inactivation of enzymes in fresh sake.     

Using computational fluid dynamics software to estimate circulation time distributions in bioreactors

Nonideal mixing in many fermentation processes can lead to concentration gradients in nutrients, oxygen, and pH, among others. These gradients are likely to influence cellular behavior, growth, or yield of the fermentation process. Frequency of exposure to these gradients can be defined by the circulation time distribution (CTD). There are few examples of CTDs in the literature, and experimental determination of CTD is at best a challenging task. The goal in this study was to determine whether computational fluid dynamics (CFD) software (FLUENT 4 and MixSim) could be used to characterize the CTD in a single-impeller mixing tank. To accomplish this, CFD software was used to simulate flow fields in three different mixing tanks by meshing the tanks with a grid of elements and solving the Navier-Stokes equations using the kappa-epsilon turbulence model. Tracer particles were released from a reference zone within the simulated flow fields, particle trajectories were simulated for 30 s, and the time taken for these tracer particles to return to the reference zone was calculated. CTDs determined by experimental measurement, which showed distinct features (log-normal, bimodal, and unimodal), were compared with CTDs determined using CFD simulation. Reproducing the signal processing procedures used in each of the experiments, CFD simulations captured the characteristic features of the experimentally measured CTDs. The CFD data suggests new signal processing procedures that predict unimodal CTDs for all three tanks.     

低温条件下常用环境消毒剂对新冠病毒灭活效果的CFD模拟和分析

新冠肺炎疫情以来,冷链食品外包装上的新冠病毒频繁引起局部疫情,对现有的环境消杀体系带来了巨大挑战。为了探究低温环境消毒剂灭活效果下降的内在机制,并为冷链过程中合理选用消毒剂提供依据,对低温条件下消毒剂的灭活过程进行计算流体力学(CFD)建模。以环境消杀常用的75%乙醇溶液和次氯酸钠溶液为工作体系,探讨了不同温度下两种溶液对新冠病毒灭活特性的影响规律。结果表明,75%乙醇溶液的灭活特性在低温环境下受温度的影响较小,而次氯酸钠溶液在环境温度为253.15 K时会发生凝固现象使得其灭活效率显著降低。该模型的模拟结果与实验结果趋势一致,可为冷链运输中疫情防控提供参考。     

Microfluidic reactor for continuous cultivation of Saccharomyces cerevisiae

A diffusion-based microreactor system operated with a reaction volume of 8 μL is presented and characterized to intensify the process understanding in microscale cultivations. Its potential as screening tool for biological processes is evaluated. The advantage of the designed microbioreactor is the use for the continuous cultivation mode by integrating online measurement technique for dissolved oxygen (DO) and optical density (OD). A further advantage is the broaden application for biological systems. The bioreactor geometry was chosen to achieve homogeneous flow during continuous process operation. The device consisted of a microstructured top layer made of poly(dimethylsiloxane) (PDMS), which was designed and fabricated using UV-depth and soft lithography assembled with a glass bottom. CFD simulation data used for geometry design were verified via microparticle-image-velocimetry (μPIV). In the used microreactor geometry no concentration gradients occurred along the entire reaction volume because of rapid diffusive mixing, the homogeneous medium flow inside the growth chamber of the microreactor could be realized. Undesirable bubble formation before and during operation was reduced by using degassed medium as well as moistened and moderate incident air flow above the gas permeable PDMS membrane. Because of this a passive oxygen supply of the culture medium in the device is ensured by diffusion through the PDMS membrane. The oxygen supply itself was monitored online via integrated DO sensors based on a fluorescent dye complex. An adequate overall volumetric oxygen transfer coefficient K(L)a as well as mechanical stability of the device were accomplished for a membrane thickness of 300 μm. Experimental investigations considering measurements of OD (online) and several metabolite concentrations (offline) in a modified Verduyn medium. The used model organism Saccharomyces cerevisiae DSM 2155 tended to strong reactor wall growth resembling a biofilm.     

Role of cytoplasmic vacuoles in varicella-zoster virus glycoprotein trafficking and virion envelopment.

Varicella-zoster virus (VZV) encodes several glycoproteins which are present on both mature viral envelopes and the surfaces of infected cell membranes. Mechanisms of VZV glycoprotein transport and virion envelopment were investigated by both continuous radiolabeling and pulse-chase analyses with tritiated fucose in VZV-infected cells. We studied in detail the large cytoplasmic vacuoles which were present in infected cells but absent from uninfected cells. The specific activity in each subcellular compartment was defined by quantitative electron microscope autoradiography, using a cross-fire probability matrix analysis to more accurately assess the individual compartment demarcated by the silver grains. By these techniques, we documented a progression of activity originating in the Golgi apparatus and traveling through the post-Golgi region into virus-induced cytoplasmic vacuoles and finally to areas of the cellular membrane associated with the egress of viral particles. Significant amounts of radiolabel were not observed in the nucleus, and only low levels of radiolabel were associated with the cellular membrane not involved with the egress of viral particles. In addition, immunolabeling of Lowicryl-embedded VZV-infected cells demonstrated the presence of VZV glycoproteins within cytoplasmic vacuole membranes as well as on virion envelopes. These observations suggested that cytoplasmic vacuoles harbored VZV-specified glycoproteins and were also the predominant site of VZV virion envelopment within the infected cell. Neither enveloped nor unenveloped viral particles were observed within the Golgi apparatus itself.     

Viral clearance in end-to-end integrated continuous process for mAb purification: Total flow-through integrated polishing on two columns connected to virus filtration

There are few reports of the adoption of continuous processes in bioproduction, particularly the implementation of end-to-end continuous or integrated processes, due to difficulties such as feed adjustment and incorporating virus filtration. Here, we propose an end-to-end integrated continuous process for a monoclonal antibody (mAb) with three integrated process segments: upstream production processes with pool-less direct connection, pooled low pH virus inactivation with pH control and a total flow-through integrated polishing process in which two columns were directly connected with a virus filter. The pooled virus inactivation step defines the batch, and high impurities reduction and mAb recovery were achieved for batches conducted in succession. Viral clearance tests also confirmed robust virus reduction for the flow-through two-column chromatography and the virus filtration steps. Additionally, viral clearance tests with two different hollow fiber virus filters operated at flux ranging from 1.5 to 40 LMH (liters per effective surface area of filter in square meters per hour) confirmed robust virus reduction over these ranges. Complete clearance with virus logarithmic reduction value ≥4 was achieved even with a process pause at the lowest flux. The end-to-end integrated continuous process proposed in this study is amenable to production processes, and the investigated virus filters have excellent applicability to continuous processes conducted at constant flux.     

Modelling and optimization of inactivation of Lactobacillus plantarum by pulsed electric field treatment

AIMS: The effect of critical pulsed electric field (PEF) process parameters, such as electric field strength, pulse length and number of pulses, on inactivation of Lactobacillus plantarum was investigated. METHODS AND RESULTS: Experiments were performed in a pH 4.5 sodium phosphate buffer having a conductivity of 0.1 S m-1, using a laboratory-scale continuous PEF apparatus with a co-linear treatment chamber. An inactivation model was developed as a function of field strength, pulse length and number of pulses. Based on this inactivation model, the conditions for a PEF treatment were optimized with respect to the minimum energy required to obtain a certain level of inactivation. It was shown that the least efficient process parameter in the range investigated was the number of pulses. The most efficient way to optimize inactivation of Lact. plantarum was to increase the field strength up to 25.7 kV cm-1, at the shortest pulse length investigated, 0.85 micros, and using a minimum number of pulses. The highest inactivation of Lact. plantarum at the lowest energy costs is obtained by using the equation: E=26.7tau0.23, in which E is the field strength and tau the pulse length. An optimum is reached by substituting tau with 5.1. CONCLUSIONS: This study demonstrates that the correct choice of parameters, as predicted by the model described here, can considerably improve the PEF process. SIGNIFICANCE AND IMPACT OF THE STUDY: The knowledge gained in this study improves the understanding of the limitations and opportunities of the PEF process. Consequently, the advantage of the PEF process as a new option for non-thermal decontamination can be better utilized.     

Numerical investigation of haemodynamics in a helical-type artery bypass graft using non-Newtonian multiphase model

The classic single-phase Newtonian blood flow model ignores the motion of red blood cells (RBCs) and their interaction with plasma. To address these issues, we adopted a multiphase non-Newtonian model to carry out a comparative study between a helical artery bypass graft (ABG) and a conventional ABG in which the blood flow is composed of plasma and RBCs. The investigation focused on the mechanism of RBC buildup in an ABG but the haemodynamic parameters obtained by single-phase and multiphase models were also compared. The aggregation of RBCs along the inside wall of a conventional ABG and at the heel of its distal anastomosis was predicted while a poor aggregation was observed along the helical ABG. In addition, RBCs were observed to gradually sediment along the gravity direction. However, the computed haemodynamic parameters by multiphase model qualitatively agreed well with those by single-phase model. It was concluded that (1) the single-phase computational fluid dynamics (CFD) is reasonable to do the computation of haemodynamic parameters in ABGs; (2) secondary flow does not definitely produce buildup of RBCs in the inside curvature, its configuration played an important role in the movement of RBCs and the dominating one-way rotating flow in a helical ABG guaranteed no buildup of RBCs on its inside wall and (3) gravity direction is important for the movement of RBCs which may help to explain why doing exercise is good for human health. This study helps to shed light on the migration of RBCs in ABGs, which cannot be explored by single-phase CFD models, and provides more understanding of the underlying flow mechanism for ABG failure.     

Computational prediction of blend time in a large-scale viral inactivation process for monoclonal antibodies biomanufacturing

Viral inactivation (VI) is a process widely used across the pharmaceutical industry to eliminate the cytotoxicity resulting from trace levels of viruses introduced by adventitious agents. This process requires adding Triton X-100, a non-ionic detergent solution, to the protein solution and allowing sufficient time for this agent to inactivate the viruses. Differences in process parameters associated with vessel designs, aeration rate, and many other physical attributes can introduce variability in the process, thus making predicting the required blending time to achieve the desired homogeneity of Triton X-100 more critical and complex. In this study we utilized a CFD model based on the lattice Boltzmann method (LBM) to predict the blend time to homogenize a Triton X-100 solution added during a typical full-scale commercial VI process in a vessel equipped with an HE-3-impeller for different modalities of the Triton X-100 addition (batch vs. continuous). Although direct experimental progress of the blending process was not possible because of GMP restrictions, the degree of homogeneity measured at the end of the process confirmed that Triton X-100 was appropriately dispersed, as required, and as computationally predicted here. The results obtained in this study were used to support actual production at the biomanufacturing site.     

Monte Carlo simulation of bifurcation in the intracellular viral kinetics

Intracellular viral kinetics are of special interest because the virion population inside an infected cell is well known to tend to grow exponentially and the corresponding kinetics may be unstable. To clarify the special features of such kinetics, we present Monte Carlo simulations taking into account the key steps of virion formation and competition of the host and viral mRNA for the host translation apparatus. Asymptotically, the model employed predicts either a stable steady state or 'ignition' with the unlimited viral growth. Under steady-state conditions, the mean square fluctuations of the viral genome and virion numbers are found to be appreciably larger than those expected on the basis of the Poissonian distribution. In the case of unstable kinetics, the simulations show the type of deviations from the corresponding mean-field results.     

A CFD model for predicting protein aggregation in low-pH virial inactivation for mAb production

Significant amounts of soluble product aggregates were observed in the low-pH viral inactivation (VI) operation during an initial scale-up run for an immunoglobulin-G 4 (IgG4) monoclonal antibody (mAb IgG4-N1). Being earlier in development, a scale-down model did not exist, nor was it practical to use costly Protein A eluate (PAE) for testing the VI process at scale, thus, a computational fluid dynamics (CFD)-based high-molecular weight (HMW) prediction model was developed for troubleshooting and risk mitigation. It was previously reported that the IgG4-N1 molecules upon exposure to low pH tend to change into transient and partially unfolded monomers during VI acidification (i.e., VIA) and form aggregates after neutralization (i.e., VIN). Therefore, the CFD model reported here focuses on the VIA step. The model mimics the continuous addition of acid to PAE and tracks acid distribution during VIA. Based on the simulated low-pH zone (≤pH 3.3) profiles and PAE properties, the integrated low-pH zone (ILPZ) value was obtained to predict HMW level at the VI step. The simulations were performed to examine the operating parameters, such as agitation speed, acid addition rate, and protein concentration of PAE, of the pilot scale (50–200 L) runs. The conditions with predictions of no product aggregation risk were recommended to the real scale-up runs, resulted in 100% success rate of the consecutive 12 pilot-scale runs. This study demonstrated that the CFD-based HMW prediction model could be used as a tool to facilitate the scale up of the low-pH VI process directly from bench to pilot/production scale.