Implementation of chemometrics,design of experiments,and neural network analysis for prior process knowledge assessment,failure modes and effect analysis,scale-down model development,and process characterization for a chromatographic purification of Teriparatide

Process understanding and characterization forms the foundation, ensuring consistent and robust biologics manufacturing process. Using appropriate modeling tools and machine learning approaches, the process data can be monitored in real time to avoid manufacturing risks. In this article, we have outlined an approach toward implementation of chemometrics and machine learning tools (neural network analysis) to model and predict the behavior of a mixed-mode chromatography step for a biosimilar (Teriparatide) as a case study. The process development data and process knowledge was assimilated into a prior process knowledge assessment using chemometrics tools to derive important parameters critical to performance indicators (i.e., potential quality and process attributes) and to establish the severity ranking for the FMEA analysis. The characterization data of the chromatographic operation are presented alongwith the determination of the critical, key and non- key process parameters, set points, operating, process acceptance and characterized ranges. The scale-down model establishment was assessed using traditional approaches and novel approaches like batch evolution model and neural network analysis. The batch evolution model was further used to demonstrate batch monitoring through direct chromatographic data, thus demonstrating its application for continuos process verification. Assimilation of process knowledge through a structured data acquisition approach, built-in from process development to continuous process verification was demonstrated to result in a data analytics driven model that can be coupled with machine learning tools for real time process monitoring. We recommend application of these approaches with the FDA guidance on stage wise process development and validation to reduce manufacturing risks.     

Conversion of a CHO cell culture process from perfusion to fed-batch technology without altering product quality

During the development of a new drug product, it is a common strategy to develop a first-generation process with the aim to rapidly produce material for pre-clinical and early stage clinical trials. At a later stage of the development, a second-generation process is then introduced with the aim to supply late-stage clinical trials as well as market needs. This work was aimed at comparing the performance of two different CHO cell culture processes (perfusion and fed-batch) used for the production of a therapeutically active recombinant glycoprotein at industrial pilot-scale. The first-generation process was based on the Fibra-Cel packed-bed perfusion technology. It appeared during the development of the candidate drug that high therapeutic doses were required (>100mg per dose), and that future market demand would exceed 100 kg per year. This exceeded by far the production capacity of the first-generation process, and triggered a change of technology from a packed-bed perfusion process with limited scale-up capabilities to a fed-batch process with scale-up potential to typical bioreactor sizes of 15m(3) or more. The productivity per bioreactor unit volume (in product m(-3)year(-1)) of the fed-batch process was about 70% of the level reached with the first-generation perfusion process. However, since the packed-bed perfusion system was limited in scale (0.6m(3) maximum) compared to the volumes reached in suspension cultures (15m(3)), the fed-batch was selected as second-generation process. In fact, the overall process performance (in product year(-1)) was about 18-fold higher for the fed-batch compared to the perfusion mode. Data from perfusion and fed-batch harvests samples indicated that comparable product quality (relative abundance of monomers dimers and aggregates; N-glycan sialylation level; isoforms distribution) was obtained in both processes. To further confirm this observation, purification to homogeneity of the harvest material from both processes, followed by a complementary set of studies (e.g. full physico-chemical characterization, assessment of in vitro and in vivo bioactivity, comparative pharmacokinetics and pharmacodynamics studies in relevant species, etc.) would be required. Finally, this illustrates the need to fix the production process early during the development of a new drug product in order to minimize process conversion efforts and to shorten product development time lines.     

A quality assurance initiative for commercial-scale production in high-throughput cryopreservation of blue catfish sperm

Cryopreservation of fish sperm has been studied for decades at a laboratory (research) scale. However, high-throughput cryopreservation of fish sperm has recently been developed to enable industrial-scale production. This study treated blue catfish (Ictalurus furcatus) sperm high-throughput cryopreservation as a manufacturing production line and initiated quality assurance plan development. The main objectives were to identify: (1) the main production quality characteristics; (2) the process features for quality assurance; (3) the internal quality characteristics and their specification designs; (4) the quality control and process capability evaluation methods, and (5) the directions for further improvements and applications. The essential product quality characteristics were identified as fertility-related characteristics. Specification design which established the tolerance levels according to demand and process constraints was performed based on these quality characteristics. Meanwhile, to ensure integrity throughout the process, internal quality characteristics (characteristics at each quality control point within process) that could affect fertility-related quality characteristics were defined with specifications. Due to the process feature of 100% inspection (quality inspection of every fish), a specific calculation method, use of cumulative sum (CUSUM) control charts, was applied to monitor each quality characteristic. An index of overall process evaluation, process capacity, was analyzed based on in-control process and the designed specifications, which further integrates the quality assurance plan. With the established quality assurance plan, the process could operate stably and quality of products would be reliable.     

Enhancement of hydrolysis and acidification of solid organic waste by a rotational drum fermentation system with methanogenic leachate recirculation

A cascade process of a rotational drum fermentation system with leachate recirculation from a methanogenic to the acidogenic reactor was constructed to enhance the hydrolysis and acidification of solid organic waste. Using fresh soybean meal as substrates, two process configurations, Cascade process 1 and 2, without and with leachate recirculation, respectively were employed to perform the experimental estimation under mesophilic condition and a total HRT of 20 days. An apparent first-order hydrolysis rate constant of 9.0 x 10(-3)/d for Cascade process 1 at pH 4.5-4.6, and 15.8 x 10(-3)/d for Cascade process 2 at pH 4.6-5.2 were obtained. The apparent VS degradation ratios ranged from 16.5% to 21.1% and total VA (as acetic acid) from 14.5 to 16.7 g/L. Occupying ratios for ionized VA decreased from 40.5% to 35.3% for Cascade process 1 and increased to 68.5% for Cascade process 2. However, occupying ratios of acetic acid decreased from 96.1% to 94.3% for Cascade process 1 and to 72.6% for Cascade process 2, whereas propionic acid and butyric acid ratios increased in acidogenesis of Cascade process 2. The leachate recirculation promoted hydrolysis of substrate in Cascade process 2, where apparent hydrolysis rate constant and VS degradation ratio were higher than that of Cascade process 1.     

Allocation of Process Gases Generated from Integrated Steelworks by an Improved System Expansion Method (7 pp)

Goal, Scope and Background Although a large number of life cycle inventory (LCI) analyses for steel-making processes or steel products have been conducted, the allocation of process gases generated from the steelworks has not yet been clearly solved. The most consistent settlement for avoiding the allocation problem has been generally known as a system expansion method. However, the existing subtracted operations for the process gases in that method are inconsistent to a system in which those gases are consumed at their unbalanced consumption ratio. The goal of this study is to suggest a more reasonable substitute for the process gases in the system expansion method and a modified system expansion method resettling the amount of process gases used. Methods To seek a more suitable one as a substituted operation of the process gas, a kind of by-product gas, in the system expansion method, it is necessary to analyze the composition of whole fuel consumed within a steelworks. Because the steelworks is supplied with a gap of electricity from the national grid electricity other than home power plants, we should also consider various carbon fossil fuels consumed in the external electric-power production. From this procedure, a composite fuel, which is composed of coal, heavy fuel oil and LNG, is derived as the alternative of the process gases such as BFG, COG, CFG, and LDG. In the sequential manufacturing line, IO(gas), which is the ratio of the quantity used to the quantity produced of each process gas, is increased as a functional unit proceeds to a following steel product of the next process. In the LCI system, where IO(gas) is higher than one, the IO(gas) is adjusted to nearly one. The adjustment of IO is conducted in the order of the amount of process gas used in the whole steelworks on the basis of the functional unit. Results and Discussion LCI analyses were carried out focusing on the alternative of the process gases for five steel products. As a functional unit goes down a lower stage, IO(gas) is increased due to the high consumption of those gases. We found a phenomenon that IO(gas) had a critical influence on the LCI results between no allocation and system expansion for the process gases through sensitivity analysis. To reduce this influence and adjust for the real situation of IO(gas), we applied an improved system expansion method to the process gases. That is, we partly substituted a process gas with LNG and rearranged the ratio between internal and external electricity (RIEE) as close as the values of IO(gas) to one. Conclusion As the alternative fuel for the process gases, a composite fuel was derived in the system expansion method. In addition to the composite fuel, which consisted of coal, HFO and LNG, an improved system expansion method was revised by adjusting a modified IO(gas) to nearly one, not the high value of IO(gas) for each process gas. Recommendation and Outlook This improved system expansion method can be applicable in the chemical industry as well as the steel industry, which have multi-function systems. Optimal LCI analysis may be achieved through the redistribution and optimization in the usage of process gases.     

Direct process integration of extraction and expanded bed adsorption in the recovery of crocetin derivatives from Fructus Gardenia

A process that integrated an extraction tank (EXT) and an expanded bed adsorption (EBA) into a new system EXT-EBA for direct purifying crocetin derivatives from Fructus Gardenia was described. Conditions were set to allow the extraction and purification in a single step. A comparison between the integrated process and the conventional process to purify crocetin derivatives was presented. The integrated process resulted in 52.79% recovery of crocin compared to 24.12% in the conventional process. The process time and solvent used were decreased in the integrated process. The result suggests that the EXT-EBA integrates extraction, clarification, and purification in a single step, greatly simplifying the process flow and reducing the cost and time of extraction and purification of crocetin derivatives from Fructus Gardenia.     

Efficient optimization of time-varying inputs in a fed-batch cell culture process using design of dynamic experiments

In cell culture process development, we rely largely on an iterative, one-factor-at-a-time procedure based on experiments that explore a limited process space. Design of experiments (DoE) addresses this issue by allowing us to analyze the effects of process inputs on process responses systematically and efficiently. However, DoE cannot be applied directly to study time-varying process inputs unless an impractically large number of bioreactors is used. Here, we adopt the methodology of design of dynamic experiments (DoDE) and incorporate dynamic feeding profiles efficiently in late-stage process development of the manufacture of therapeutic monoclonal antibodies. We found that, for the specific cell line used in this article, (1) not only can we estimate the effect of nutrient feed amount on various product attributes, but we can also estimate the effect, develop a statistical model, and use the model to optimize the slope of time-trended feed rates; (2) in addition to the slope, higher-order dynamic characteristics of time-trended feed rates can be incorporated in the design but do not have any significant effect on the responses we measured. Based on the DoDE data, we developed a statistical model and used the model to optimize several process conditions. Our effort resulted in a tangible improvement in productivity—compared with the baseline process without dynamic feeding, this optimized process in a 200-L batch achieved a 27% increase in titer and > 92% viability. We anticipate our application of DoDE to be a starting point for more efficient workflows to optimize dynamic process conditions in process development.     

Dynamics of water in supercooled aqueous solutions of glucose and poly(ethylene glycol)s as studied by dielectric spectroscopy

The dielectric behaviour of aqueous solutions of glucose, poly(ethylene glycol)s (PEGs) 200 and 600, and poly(vinyl pyrrolidone) (PVP) has been examined at different concentrations in the frequency range of 10(6)-10(-3) Hz by dielectric spectroscopy and by using differential scanning calorimetry down to 77 K from room temperature. The shape of the relaxation spectra and the temperature dependence of the relaxation rates have been critically examined along with temperature dependence of dielectric strength. In addition to the so-called primary (alpha-) relaxation process, which is responsible for the glass-transition event at T(g), another relaxation process of comparable magnitude has been found to bifurcate from the main relaxation process on the water-rich side, which continues to the sub-T(g) region, exhibiting relaxation at low frequencies. The sub-T(g) process dominates the dielectric measurements in aqueous solutions of higher PEGs, and the main relaxation process is seen as a weak process. The sub-T(g) process was not observed when water was replaced by methanol in the binary mixtures. These observations suggest that the sub-T(g) process in the aqueous mixtures is due to the reorientational motion of the 'confined' water molecules. The corresponding dielectric strength shows a noticeable change at T(g), indicating a hindered rotation of water molecules in the glassy phase. The nature of this confined water appears to be anomalous compared to most other supercooled confined liquids.     

Economic assessment of batch biodiesel production processes using homogeneous and heterogeneous alkali catalysts

An economic feasibility study on four batch processes for the production of biodiesel ranging from 1452 tonnes/year (5000 l/day) to 14,520 tonnes/year (50,000 l/day) is conducted. The four processes assessed are the (1) KOH-W process, characterized by a homogeneous KOH catalyst and hot water purification process; (2) KOH-D process, characterized by a homogeneous KOH catalyst and vacuum FAME distillation process; (3) CaO-W process, characterized by a heterogeneous CaO catalyst and hot water purification process; and (4) CaO-D process, characterized by a heterogeneous CaO catalyst and vacuum FAME distillation process. The costs of the waste cooking oil, fixed costs, and manufacturing costs for producing 7260 tonnes/year (25,000 l/day) of biodiesel by means of the four processes are estimated to be $248–256, $194–232, and $584–641 per tonne of biodiesel, respectively. Among the four processes, the manufacturing costs involved in the CaO-W process are the lowest, in the range from 1452 tonnes/year to 14,520 tonnes/year.     

Visualizing integrated bioprocess designs through "windows of operation".

This paper demonstrates a simple graphical approach for the design and analysis of a bioprocess flowsheet in which process interactions are significant. Results are presented showing how the feasible space for operation can be simulated and used both to address key design and operating decisions and to identify suitable trade-offs between operating variables, such as fermentation growth rate and disruption conditions, in order to achieve prespecified levels of process performance. Using verified models to describe the production and isolation of an intracellular protein alcohol dehydrogenase (ADH) in yeast as a test bed, a series of so-called "windows of operation" are developed at growth rates in the range of 0.06-0.28 h(-1) and for a range of overall process specifications. The effects of altering the process design performance specification as defined by the level of cell debris removal and the overall process productivity on the size and position of the feasible space were investigated to demonstrate the sensitivity of the flowsheet to changes in process objectives. Using the approach it has been possible to visualise the processing trade-offs required to increase performance in terms of the level of cell debris removal by 50% and the overall process productivity by 400% from a defined base level. The approach provides a convenient tool when designing integrated bioprocesses by enabling process options to be compared visually and can help in achieving better process designs and accelerating process development for the biological process industry.     

Expanded bed adsorption in the purification of monoclonal antibodies: a comparison of process alternatives

Expanded bed adsorption (EBA) was examined as the initial capture/purification step in the purification of monoclonal antibodies from Chinese hamster ovary (CHO) cultures. Two process alternatives each using EBA were compared to a conventional Protein A process without EBA. One alternative used Protein A affinity EBA followed by packed-bed cation and anion-exchange steps. The other alternative used cation-exchange EBA as the capture step followed by packed-bed Protein A and anion-exchange steps. The process using Protein A EBA produced comparable purity (host cell protein, DNA, Protein A, antibody aggregate) to the conventional process. However, the Protein A EBA column showed a significant decrease in dynamic capacity with a limited number of cycles. The process using cation EBA achieved comparable levels of host cell proteins (HCP) and DNA but not antibody aggregate or leached Protein A compared to the conventional process.     

XCSc: a novel approach to clustering with extended classifier system

In this paper, we propose a novel approach to clustering noisy and complex data sets based on the eXtend Classifier Systems (XCS). The proposed approach, termed XCSc, has three main processes: (a) a learning process to evolve the rule population, (b) a rule compacting process to remove redundant rules after the learning process, and (c) a rule merging process to deal with the overlapping rules that commonly occur between the clusters. In the first process, we have modified the clustering mechanisms of the current available XCS and developed a new accelerate learning method to improve the quality of the evolved rule population. In the second process, an effective rule compacting algorithm is utilized. The rule merging process is based on our newly proposed agglomerative hierarchical rule merging algorithm, which comprises the following steps: (i) all the generated rules are modeled by a graph, with each rule representing a node; (ii) the vertices in the graph are merged to form a number of sub-graphs (i.e. rule clusters) under some pre-defined criteria, which generates the final rule set to represent the clusters; (iii) each data is re-checked and assigned to a cluster that it belongs to, guided by the final rule set. In our experiments, we compared the proposed XCSc with CHAMELEON, a benchmark algorithm well known for its excellent performance, on a number of challenging data sets. The results show that the proposed approach outperforms CHAMELEON in the successful rate, and also demonstrates good stability.     

Application of high‐throughput mini‐bioreactor system for systematic scale‐down modeling,process characterization,and control strategy development

High‐throughput systems and processes have typically been targeted for process development and optimization in the bioprocessing industry. For process characterization, bench scale bioreactors have been the system of choice. Due to the need for performing different process conditions for multiple process parameters, the process characterization studies typically span several months and are considered time and resource intensive. In this study, we have shown the application of a high‐throughput mini‐bioreactor system viz. the Advanced Microscale Bioreactor (ambr15^TM), to perform process characterization in less than a month and develop an input control strategy. As a pre‐requisite to process characterization, a scale‐down model was first developed in the ambr system (15 mL) using statistical multivariate analysis techniques that showed comparability with both manufacturing scale (15,000 L) and bench scale (5 L). Volumetric sparge rates were matched between ambr and manufacturing scale, and the ambr process matched the pCO₂ profiles as well as several other process and product quality parameters. The scale‐down model was used to perform the process characterization DoE study and product quality results were generated. Upon comparison with DoE data from the bench scale bioreactors, similar effects of process parameters on process yield and product quality were identified between the two systems. We used the ambr data for setting action limits for the critical controlled parameters (CCPs), which were comparable to those from bench scale bioreactor data. In other words, the current work shows that the ambr15^TM system is capable of replacing the bench scale bioreactor system for routine process development and process characterization. © 2015 American Institute of Chemical Engineers Biotechnol. Prog., 31:1623–1632, 2015     

Enhanced process understanding and multivariate prediction of the relationship between cell culture process and monoclonal antibody quality

This work investigates the insights and understanding which can be deduced from predictive process models for the product quality of a monoclonal antibody based on designed high‐throughput cell culture experiments performed at milliliter (ambr‐15^®) scale. The investigated process conditions include various media supplements as well as pH and temperature shifts applied during the process. First, principal component analysis (PCA) is used to show the strong correlation characteristics among the product quality attributes including aggregates, fragments, charge variants, and glycans. Then, partial least square regression (PLS1 and PLS2) is applied to predict the product quality variables based on process information (one by one or simultaneously). The comparison of those two modeling techniques shows that a single (PLS2) model is capable of revealing the interrelationship of the process characteristics to the large set product quality variables. In order to show the dynamic evolution of the process predictability separate models are defined at different time points showing that several product quality attributes are mainly driven by the media composition and, hence, can be decently predicted from early on in the process, while others are strongly affected by process parameter changes during the process. Finally, by coupling the PLS2 models with a genetic algorithm first the model performance can be further improved and, most importantly, the interpretation of the large‐dimensioned process–product‐interrelationship can be significantly simplified. The generally applicable toolset presented in this case study provides a solid basis for decision making and process optimization throughout process development. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:1368–1380, 2017     

Performance assessment of surface-guided radiation therapy and patient setup in head-and-neck and breast cancer patients based on statistical process control

PurposeTo assess the effectiveness of SGRT in clinical applications through statistical process control (SPC).MethodsTaking the patients’ positioning through optical surface imaging (OSI) as a process, the average level of process execution was defined as the process mean. Setup errors detected by cone-beam computed tomography (CBCT) and OSI were extracted for head-and-neck cancer (HNC) and breast cancer patients. These data were used to construct individual and exponentially weighted moving average (EWMA) control charts to analyze outlier fractions and small process shifts from the process mean. Using the control charts and process capability indices derived from this process, the patient positioning-related OSI performance and setup error were analyzed for each patient.ResultsOutlier fractions and small shifts from the process mean that are indicative of setup errors were found to be widely prevalent, with the outliers randomly distributed between fractions. A systematic error of up to 1.6 mm between the OSI and CBCT results was observed in all directions, indicating a significantly degraded OSI performance. Adjusting this systematic error for each patient using setup errors of the first five fractions could effectively mitigate these effects. Process capability analysis following adjustment for systematic error indicated that OSI performance was acceptable (process capability index C_pk = 1.0) for HNC patients but unacceptable (C_pk < 0.75) for breast cancer patients.ConclusionSPC is a powerful tool for detecting the outlier fractions and process changes. Our application of SPC to patient-specific evaluations validated the suitability of OSI in clinical applications involving patient positioning.     

Implementing Process Analytical Technology for the Production of Recombinant Proteins in Escherichia coli Using an Advanced Controller Scheme

The performance of a bioreactor in meeting process goals is affected by the microorganism used, medium composition, and operating conditions. A typical bioreactor uses a supervisory control and data acquisition (SCADA) system for control, and a combination of software and hardware tools for real‐time data analysis. However, when the process is disrupted by utility or instrumentation failure, typical process controllers may be unable to reinstate normal operating conditions before the cells in the reactor shift to unfavorable metabolic regimes. The objective of this study is to examine how the response of a controller affects process recovery when disruptive incidences occur under a process analytical technology (PAT) framework. The process used for this investigation is the production of lethal toxin‐neutralizing factor (LTNF) by Escherichia coli (E. coli), which is controlled by a decoupled input–output‐linearizing controller (DIOLC). The performance of the DIOLC is compared to a proportional integral derivative (PID) controller subjected to the same conditions. The disruptions are introduced manually and the effect of controller action on process recovery and LTNF synthesis is measured in terms of peak purity and concentration. It is observed that DIOLC performs better after reinstating operating conditions and results in a meaningful improvement in performance.     

Cell bank characterization and fermentation optimization for production of recombinant heavy chain C-terminal fragment of botulinum neurotoxin serotype E (rBoNTE(H(c)): antigen E) by Pichia pastoris

A process was developed for production of a candidate vaccine antigen, recombinant C-terminal heavy chain fragment of the botulinum neurotoxin serotype E, rBoNTE(H(c)) in Pichia pastoris. P. pastoris strain GS115 was transformed with the rBoNTE(H(c)) gene inserted into pHILD4 Escherichia coli-P. pastoris shuttle plasmid. The clone was characterized for genetic stability, copy number, and BoNTE(H(c)) sequence. Expression of rBoNTE(H(c)) from the Mut(+) HIS4 clone was confirmed in the shake-flask, prior to developing a fed-batch fermentation process at 5 and 19 L scale. The fermentation process consists of a glycerol growth phase in batch and fed-batch mode using a defined medium followed by a glycerol/methanol transition phase for adaptation to growth on methanol and a methanol induction phase resulting in the production of rBoNTE(H(c)). Specific growth rate, ratio of growth to induction phase, and time of induction were critical for optimal rBoNTE(H(c)) production and minimal proteolytic degradation. A computer-controlled exponential growth model was used for process automation and off-gas analysis was used for process monitoring. The optimized process had an induction time of 9 h on methanol and produced up to 3 mg of rBoNTE(H(c)) per gram wet cell mass as determined by HPLC and Western blot analysis.     

Anisotropic viscoelastic properties of cortical bone

Relaxation Young's modulus of cortical bone was investigated for two different directions with respect to the longitudinal axis of bone (bone axis, BA): the modulus parallel (P) and normal (N) to the BA. The relaxation modulus was analyzed by fitting to the empirical equation previously proposed for cortical bones, i.e., a linear combination of two Kohlraush-Williams-Watts (KWW) functions (Iyo et al., 2003. Biorheology, submitted): E(t)=E0 (A1 exp[-(t/tau1)beta]+(1-A1) exp[-(t/tau2)gamma]), [0 < A1, beta, gamma < 1], where E0 is the initial modulus value E0. Tau1 and tau2(>tau1) are characteristic times of the relaxation, A1 is the fractional contribution of the fast relaxation (KWW1 process) to the whole relaxation process, and beta and gamma are parameters describing the shape of the relaxation modulus. In both P and N samples, the relaxation modulus was described well by the empirical equation. The KWW1 process of a P sample almost completely coincided with that of an N sample. In the slow process (KWW2 process), there was a difference between the relaxation modulus of a P sample and that of an N sample. The results indicate that the KWW1 process in the empirical equation represents the relaxation in the collagen matrix in bone and that the KWW2 process is related to a higher-order structure of bone that is responsible for the anisotropic mechanical properties of bone.     

Characterization of a Monoclonal Antibody Cell Culture Production Process Using a Quality by Design Approach

The goal of quality by design (QbD) in cell culture manufacturing is to develop manufacturing processes which deliver products with consistent critical quality attributes (CQAs). QbD approaches can lead to better process understanding through the use of process parameter risk ranking and statistical design of experiments (DOE). The QbD process starts with an analysis of process parameter risk with respect to CQAs and key performance indicators (KPIs). Initial DOE study designs and their factor test ranges are based on the outcomes of the process parameter risk ranking exercises. Initial DOE studies screen factors for significant influences on CQAs as well as characterize responses for process KPIs. In the case study provided here, multifactor process characterization studies using a scale-down model resulted in significant variation in charge heterogeneity of a monoclonal antibody (MAb) as measured by ion-exchange chromatography (IEC). Iterative DOE studies, using both screening and response surface designs, were used to narrow the operating parameter ranges so that charge heterogeneity could be controlled to an acceptable level. The data from the DOE studies were used to predict worst-case conditions, which were then verified by testing at those conditions. Using the approach described here, multivariate process parameter ranges were identified that yield acceptable CQA levels and that still provide operational flexibility for manufacturing.