Publications

Authors: Nirupaplava Metta, Marianthi, Ierapetritou, Rohit, Ramachandran

Paper Link: Link

Abstract: The population balance approach (PBM) is generally used in the literature to simulate a milling process. The formulation of a breakage kernel to represent particle breakage phenomenon is an important part of the model. This study proposes a methodology to estimate parameters of a breakage kernel that captures material property dependent particle level dynamics through discrete element method (DEM) simulations of a comill process. The DEM model takes into account a threshold impact energy that if exceeded, results in granule breakage. The impact energy distribution data for various size classes and impellor speeds is obtained from DEM. Comill experiments at various impeller speeds result in different observed size distributions and other process variables such as hold up amount, and time required for process to reach steady state. An iterative algorithm is proposed that uses mechanistic information from DEM and process variables from experiments to calibrate the breakage kernel through which material specific kernel parameters are estimated. A multi-scale modeling framework utilizing DEM, PBM as well as experimental data is developed. The framework is implemented to estimate material specific properties using milling experimental data at various impeller speeds. The milled particle size distribution predicted from the model with parameters estimated using this framework, demonstrated excellent agreement with experimental results.

Authors: Sarang Oka, David Smrčka, Anjali Kataria, Heather Emadym Fernando Muzzio, František Štěpánek, Rohit Ramachandran

Paper Link: Link

Abstract: In this study, the origins of granule content non-uniformity in the high-shear wet granulation of a model two-component pharmaceutical blend were investigated. Using acetaminophen as the active pharmaceutical ingredient (API) and microcrystalline cellulose as the excipient, the distribution of the API across the granule size classes was measured for a range of conditions that differed in the duration of the initial dry mixing stage, the overall composition of the blend and the wet massing time. The coarse granule fractions were found to be systematically sub-potent, while the fines were enriched in the API. The extent of content non-uniformity was found to be dependent on two factors – powder segregation during dry mixing and redistribution of the API between the granule size fractions during the wet massing phase. The latter was demonstrated in an experiment where the excipient was pre-granulated, the API was added later and wet massed. The content non-uniformity in this case was comparable to that obtained when both components were present in the granulator from the beginning. With increasing wet massing time, the extent of content non-uniformity decreased, indicating that longer wet massing times might be a solution for systems with a natural tendency for component segregation.

Authors: Wei Meng, Sarang Oka, Xue Liu, Thamer Omer, Rohit Ramachandran & Fernando J. Muzzio

Paper Link: Link

Abstract: Wet granulation is widely used in the pharmaceutical industry. This advantageous technology is capable of enhancing compression and powder handling, decreasing ingredient segregation, and promoting blend and content uniformity. Currently, a high level of interest exists in the continuous version of this technology, both by the US Food and Drug Administration (FDA), and by pharmaceutical manufacturers.

Authors: Maitraye Sen, Subhodh Karkala, Savitha Panikar, Olav Lyngberg, Mark Johnson, Alexander Marchut, Elisäbeth Schäfer, and Rohit Ramachandran

Paper Link: Link

Abstract: A discrete element model (DEM) has been developed for an industrial batch bin blender in which three different types of materials are mixed. The mixing dynamics have been evaluated from a model-based study with respect to the blend critical quality attributes (CQAs) which are relative standard deviation (RSD) and segregation intensity. In the actual industrial setup, a sensor mounted on the blender lid is used to determine the blend composition in this region. A model-based analysis has been used to understand the mixing efficiency in the other zones inside the blender and to determine if the data obtained near the blender-lid region are able to provide a good representation of the overall blend quality. Sub-optimal mixing zones have been identified and other potential sampling locations have been investigated in order to obtain a good approximation of the blend variability. The model has been used to study how the mixing efficiency can be improved by varying the key processing parameters, i.e., blender RPM/speed, fill level/volume and loading order. Both segregation intensity and RSD reduce at a lower fill level and higher blender RPM and are a function of the mixing time. This work demonstrates the use of a model-based approach to improve process knowledge regarding a pharmaceutical mixing process. The model can be used to acquire qualitative information about the influence of different critical process parameters and equipment geometry on the mixing dynamics.

Authors: Anik Chaturbedi, Chandra Kanth Bandi, Dheeraj Reddy, Preetanshu Pandey, Ajit Narang, Dilbir Bindra, Li Tao, Junshu Zhao, Jinjiang Li, Munir Hussain, Rohit Ramachandran

Paper Link: Link

Abstract: Population balance models (PBM) have been used traditionally to model high shear wet granulation (HSWG) with wet binder addition where the binder is pre-dissolved in a liquid and added to the granulator. However, wet granulation with dry binder addition can not accurately be modeled with the models developed for wet binder addition since it involves the additional step of dissolution of the dry binder present in the granulator in the pure liquid added during liquid addition stage. In this work, a reduced order multi-compartment population balance model integrated with binder dissolution model was developed to address the differences in average diameter of particles obtained from dry and wet binder addition. Experimental data were generated on a 10-L PMA granulator using wet and dry binder addition modes. The experimental data were used to estimate the model tuning parameters to validate the model which was further used as predictive tool. This model showed good agreement with experimental data in capturing the trends in average particle diameter for two different binders, hydroxypropyl cellulose (HPC) and polyvinylpyrrolidone (PVP). The model was also able to accurately predict the average diameter for both the wet binder and dry binder addition cases.

Authors: Franklin E.Bettencourt, Anik Chaturbedi, and Rohit Ramachandran

Paper Link: Link

Abstract: In order to solve high resolution PBMs to simulate real systems, with high accuracy and speed, a comprehensive and robust parallelization framework is needed. In this work, parallelization using just Message Passing Interface (MPI) and a more advanced method using a hybrid MPI + OpenMP (Open Multi-Processing) technique, have been applied to simulate high resolution PBMs on the computing clusters, SOEHPC and Stampede. We study the speed up and the scale up of these parallelization techniques for different system sizes and different computer architectures to come up with one of the fastest ways to solve a PBM to date. Parallel PBMs ran approximately 50–60 times faster, when using 128 cores, than the serial PBMs ran. In this work it is found that hybrid MPI + OMP methods which account for socket affinities led to the fastest PBM compute times and about 80% less memory than a purely MPI approach.