Publications

Authors: Nirupaplava Metta, Rohit Ramachandran, Marianthi Ierapetritou

Paper Link: Link

Abstract: Identification of feasible region of operations in multivariate processes is a problem of interest in several fields. This is particularly challenging when the process model is black-box in nature and/or is computationally expensive, as analytical solutions are not available and the number of possible model evaluations is limited. An efficient methodology is required to identify samples where the model is evaluated for developing a computationally efficient surrogate model. In this work, an artificial neural network based surrogate model is proposed which is integrated with a statistical-based approach (Jack-knifing) to estimate the variance of the surrogate model prediction. This allows implementation of an adaptive sampling approach where new samples are identified close to the feasible region boundary or in regions of high prediction uncertainty. The proposed approach performs better than a previously published kriging based method for different dimensionality case studies.

Authors: AnikChaturbedi, SanyuktaPatil, RohitRamachandran, Nina C.Shapley

Paper Link: Link

Abstract: Environmentally benign biopolymer particles formed from natural carbohydrate polymers alginate and chitosan have shown great capacity for adsorption of heavy metal ions from wastewater. In this study, heteroaggregates of large alginate beads (1.8 mm) and chitosan nanoparticles (250 nm) have been used to adsorb molybdate and copper ions from single ion and mixed ion solutions in order to study the equilibrium adsorption properties. Three distinct heteroaggregation regimes have been studied in order to determine how the structure and composition affect adsorption capacity. Significantly enhanced adsorption was observed for heteroaggregates compared to the sum of their individual components, in both single ion and mixed ion solutions, indicating the presence of synergistic effects. The resulting adsorption capacity values for heteroaggregates are highly competitive with previous literature results and reach a maximum capacity of approximately 200–600 mg/g adsorption of either ion in a mixed ion solution, based on the adsorbent dry mass. In addition, synergistic and competitive effects were investigated in mixed ion solutions containing oppositely charged copper and molybdate ions, in an effort to mimic more realistic wastewater, and in contrast to typical competitive adsorption studies involving only like charged ions. Moreover, the observed interaction of both ions with heteroaggregates and their components suggested that opposite charge attraction is a more effective adsorption mechanism than complexation. Finally, the heteroaggregates represent a potential path forward for using nanoparticles in practical adsorption, where the large alginate beads act as supports for the nanoparticles and as effective adsorbents.

Authors: Yingjie Chen, Ou Yang, Chaitanya Sampat, Pooja Bhalode, Rohit Ramachandran and Marianthi Ierapetritou

Paper Link: Link

Abstract: The development and application of emerging technologies of Industry 4.0 enable the realization of digital twins (DT), which facilitates the transformation of the manufacturing sector to a more agile and intelligent one. DTs are virtual constructs of physical systems that mirror the behavior and dynamics of such physical systems. A fully developed DT consists of physical components, virtual components, and information communications between the two. Integrated DTs are being applied in various processes and product industries. Although the pharmaceutical industry has evolved recently to adopt Quality-by-Design (QbD) initiatives and is undergoing a paradigm shift of digitalization to embrace Industry 4.0, there has not been a full DT application in pharmaceutical manufacturing. Therefore, there is a critical need to examine the progress of the pharmaceutical industry towards implementing DT solutions. The aim of this narrative literature review is to give an overview of the current status of DT development and its application in pharmaceutical and biopharmaceutical manufacturing. State-of-the-art Process Analytical Technology (PAT) developments, process modeling approaches, and data integration studies are reviewed. Challenges and opportunities for future research in this field are also discussed.

Authors: InduMuthancheri, Barry Long, Kevin M.Ryan, Luis Padrela, Rohit Ramachandran

Paper Link: Link

Abstract: In this study, a two-dimensional population balance model with solvent removal kinetics has been developed to predict the dynamic behavior of carbamazepine form II crystals produced by a supercritical CO2 antisolvent batch crystallization process. The model was simulated and validated using experimental crystal size distribution data (CSD). The model was able to accurately predict the behavior of CSD with a change in process operating conditions. The model was also applied to study the time evolution of aspect ratio, average crystal length, and solute concentration in the solution. Finally, solvent removal kinetics were modeled to evaluate the solvent content and drying temperature of the drying gas during the solvent removal process. The developed mathematical model and the presented results suggest the ability of the discussed approach to make suitable model predictions, which can significantly reduce the number of experimental trials required for process design, optimization, and control.

Authors: Chaitanya Sampat, Yukteshwar Baranwal, Rohit Ramachandran

Paper Link: Link

Abstract: The solution of high-dimensional PBMs using CPUs are often computationally intractable. This study focuses on the development of a scalable algorithm to parallelize the nested loops inside the PBM via a GPU framework. The developed PBM is unique since it adapts to the size of the problem and uses the GPU cores accordingly. This algorithm was parallelized for NVIDIA® GPUs as it was written in CUDA® and C/C++. The major bottleneck of such algorithms is the communication time between the CPU and the GPU. In our studies, communication time contributed to less than 1% of the total run time and a maximum speedup of about 12 over the serial CPU code was achieved. The GPU PBM achieved a speedup of about two times compared to the PBM’s multi-core configuration on a desktop computer. The speed improvements are also reported for various CPU and GPU architectures and configurations.

Authors: Lalith Kotamarthy, Nirupaplava Metta, and Rohit Ramachandran

Paper Link: Link

Abstract: Milling is an essential unit operation used for particle size reduction in solid oral dosage manufacturing. The breakage of particles in a comil is due to the intense shear applied on the particles between impeller and the screen. Breakage also occurs due to the impact from a rotating impeller. Particles exit the mill based on their size relative to the aperture size of the screen bores. This study was set up to understand the working of the comil better. A new CPP (Critical Process Parameter), in the form of batch loading was identified. It was found that there are two different regimes (quasi static regime and impact regime) in which a comil generally operates, and the effect of the CPP’s (batch loading and impeller speed) on these regimes was studied. Knowledge of the effect of upstream operations on a particular unit operation is of significant importance, especially for pharmaceutical industry. For this reason, the effect of granulation variables such as liquid-to-solid ratio, granulator impeller speed and the amount of binder in the formulation were analyzed. Milled particle size distribution and other critical quality attributes such as bulk density, friability, and porosity were studied. Batch loading and the interaction effect of batch loading with impeller speed are significant parameters that affect the quality attributes of the mill. Predictive regression models were developed for throughput of the mill, milled product bulk density and milled product tapped density (with an R2 of 0.987, 0.953, 0.995 respectively) to enable their use in downstream process modeling.

Authors: Indu Muthancheri, Rohit Ramachandra

Paper Link: Link

Abstract: In this study, a more physics-based approach was pursued to investigate the effect of wettability differences in bi-component granule systems on critical quality attributes such as granule size distribution (GSD) and content uniformity across size fractions. Two bi-component formulations: 1) ibuprofen-USP and microcrystalline cellulose (Case-I) and; 2) APAP and microcrystalline cellulose (Case-II), were systematically selected and only their net hydrophobicity was varied as a consequence of change in percentage composition. Both case studies resulted in a similar increase in mean granule size and GSD as a function of increased net hydrophobicity; however, they exhibited key differences in terms of content uniformity across size fractions. Case-I produced content uniform granules, while Case-II produced content non-uniform granules, with the increase in percentage composition of ibuprofen and acetaminophen respectively. This difference in component distribution was studied using growth regime maps and granule image analyses. Results showed that with the change in percentage composition (or increasing the amount of the hydrophobic material), the granule growth mechanism shifted between capillary-force-dominated layering mechanism and viscous-force-dominated aggregation mechanism. It was concluded that, uniform distribution of materials observed in Case-I granules was due to strong viscous-force-dominated aggregation mechanism; whereas, non-uniform distribution of materials observed in Case-II granules was due to much weaker capillary force dominated layering mechanism. This work seeks to improve the theoretical understanding of granule growth behavior in a bi-component wet granulation system with components of differing wettabilities or hydrophobicities.

Authors: Kevin M.Moroneya, Patrick Cronin, Opeyemi A.Adeleye, Barbara E.Schaller, Matthew A.Howard, Bernardo Castro-Dominguez, Rohit Ramachandran, Gavin M.Walker

Paper Link: Link

Abstract: Roller compaction (RC) is a dry granulation technique applied to improve the flow and compressibility of drug formulations. RC implementation for high drug load formulations can be challenging due to flow issues and a high consumption of active pharmaceutical ingredient (API) for robust process development. This work addresses these challenges using process modelling for design and scale-up of an RC process on the same equipment and transfer to different equipment. A modified application of existing models incorporating a new description of mass transport in the feed screw is evaluated for guaifenesin formulations with a 90% drug loading. The model is calibrated using low-throughput data on a Vector Freund TF Mini RC and used to predict ribbon density and throughput for various process settings at high-throughput. It is found that the modelling framework can reasonably predict high-throughput behaviour on the same RC but the predictive performance decreases for transfer between equipment.

Authors: Tianxiang Gao, Arun S.S.Singaravelu, Sarang Oka, Rohit Ramachandran, František Štepánek, NikhileshChawla, Heather N.Emady

Paper Link: Link

Abstract: Single drop granule formation on a static powder bed of pharmaceutical mixtures was studied to investigate the effects of hydrophobicity and primary particle size distribution on the powder bed packing structure and the content homogeneity of active pharmaceutical ingredient (API) in granules formed. The granule formation mechanisms, drop penetration time, granule morphology and internal structure have been previously investigated in a mixture of coarse microcrystalline cellulose (MCC) and fine acetaminophen (APAP). When the APAP amount increased (decreasing particle size and increasing hydrophobicity), drop penetration time increased, formation mechanisms transitioned from Spreading to Tunneling, the granules became smaller in size, and the internal porosity of the granules decreased (Gao et al., 2018). In the current study, single drop granulation on mixtures of MCC and APAP with different particle sizes were investigated for formation mechanisms and granule morphology. Additionally, the powder bed packing structure was characterized by X-ray micro-CT and the API content uniformity was measured by UV–vis spectrometry. It was found that in the mixture made from coarse MCC and fine APAP, the internal structure became heterogeneous and there were dense aggregate regions in both the powder bed and granules from 25% APAP proportion, where the transition from Spreading to Tunneling occurs. The content uniformities of granules from fine powder beds are much more compromised (indicating a discrepancy between the actual value and theoretical value) than those from coarse powder beds. This content discrepancy becomes much larger when the APAP proportion in the powder bed is higher (above 50%). This was previously observed by other researchers (Nguyen et al., 2010) and was attributed to the preferential wetting of the ingredients. It is believed that the primary particle size of the powder bed is more significant than the hydrophobicity in affecting the formation mechanism, granule internal structure, and content uniformity.

Authors: Andrés D.Román-Ospino, Ashutosh Tamrakar, Benoît Igne, ElyseTowns Dimaso, Christian Airiau, Donald J.Clancy, Glinka Pereira, Fernando J.Muzzio, Ravendra Singh, RohitRamachandran

Paper Link: Link

Abstract: This work describes the characterization of three NIR interfaces intended to monitor a continuous granulation process. Two interfaces (i.e. a barrel interface and a rotating paddle interface) were evaluated to monitor the API concentration at the entrance of the granulator, and a third interface (i.e. an outlet interface), was evaluated to examine the quality of the resulting outlet granules. The barrel interface provided an assessment of the API concentration during the feeding process by scanning the material conveyed by the screws of the loss-in-weight feeder. The rotating paddle interface analyzed discrete amounts of powder upon exiting the feeder via the accumulation of material on the paddles. Partial Least Squares (PLS) calibration models were developed using the same powder blends for the two inlet interfaces and using the outlet granules for the outlet interface. Five independent batches were used to evaluate the prediction performance of each inlet calibration model. The outlet interface produced the lowest error of prediction due to the homogeneity of the granules. The barrel interface produced lower errors of prediction than the rotating paddle interface. However, powder density affected only the barrel interface, producing deviations in the predicted values. Therefore, powder density is a factor that should be considered in the calibration sample design for spectroscopic measurements when using this type of interface. A variographic analysis demonstrated that the continuous 1-dimensional motion in the barrel and outlet interfaces produced representative measurements of each batch during calibration and test experiments, generating a low minimum practical error (MPE)

Authors: Ashutosh Tamrakar, Alex Zheng, Patrick M.Piccione, Rohit Ramachandran

Paper Link: Link

Abstract: Agitated filter drying (AFD) is a complex physical-thermal separation process which involves isolating solutes from its mother liquor. In agro-chemical and pharmaceutical industry, filter-dryers are used for sequestering active ingredients (AIs) and key intermediates from the wet cake after the crystallization step. During the agitated drying phase, the mechanical agitation of the wet cake, implemented to enhance heat and mass transport, has been commonly observed to result in formation of undesired agglomerates that require further processing. Only relatively few experimental and computational studies of the effects of operating parameters and material properties on the drying and agglomeration growth kinetics have been described in the literature. In absence of robust predictive models, the go-to solution in order to avoid the agglomeration behavior of AIs has been to use minimal agitation which is not only suboptimal but also significantly increases the drying times.