Fernando J. Muzzio, PhD, is distinguished professor of the School of Engineering at Rutgers, The State University of New Jersey, Piscataway, NJ 08855. Dr. Muzzio is a member of 

 editorial advisory board and President of Integra Continuous Manufacturing Systems.

Fernando J. Muzzio

The application of innovation strategies often achieves a business’ competitive advantage. One definition of innovation is either doing different things or doing things differently (1). In the past few decades, creating new ways of doing business has encouraged global corporations to look at competitive scenarios that provide lower-cost advantages. The innovation strategies designed to achieve cost improvements included business orchestrations or alliances, outsourcing, supplier leverage, and the offshoring of companies’ manufacturing activities to either lower tax or lower labor cost jurisdictions (2,3).

As demonstrated by recent experience during the COVID-19 pandemic, despite the expectation of providing a company with competitive advantages, the risks of extensive offshoring likely surpass its benefits (4–6). The pandemic showed that the potential impact of commerce disruptions on the worldwide supply chain of goods sold is a significant risk element of offshoring strategies. Moreover, other supply chain disruption events such as geopolitical situations, as well as other business risks such as competencies and knowledge loss, brand damage, or deficient intellectual property protection policies of manufacturing recipient countries are significant sources of risk to weight against offshoring benefits (2,5,7). Specifically, regarding pharmaceutical products, the risks of disrupting the supply chain are much deeper than simply financial; literally, tens of millions of patients in the United States and abroad depend critically on reliable access to medicines to maintain quality of life, and in many cases, life itself.

Refocusing the innovation lenses, from offshoring to reshoring

In the light of the significant threat that a future geopolitical or pandemic event poses to the pharmaceutics supply chain, reshoring manufacturing activities is the right business decision both for business continuity and for national security reasons. The translation of both needs to a coherent and viable business model for pharmaceutical manufacturing depends upon our ability to establish a strong foundation of technology innovations. It is important to identify innovations that would support making a reshoring strategy successful, not only in the short term when supporting government policies are in place, but in the long term, when sustainability depends on profitability. Taking generic-drug pharmaceutical manufacturing as an example, the authors seek to propose a viable pharmaceutical manufacturing reshoring strategy. A significant number of generic-drug products are essential and needed medicines that most people depend on. Still, manufacturers neglect to make them on US soil due to profit margins and capacity constraints (8). If a case can be made for successfully reshoring manufacturing of generic pharmaceutical products, where profit margins per product unit are small, it is then likely that a reshoring innovation strategy can be successful for any pharmaceutical product.

COVID-19’s exacerbation of the drug shortage problem

Generic-drug manufacturing has been under close attention by the US Congress and FDA due to the importance of these drugs’ availability as a public health imperative need. The causes of generic drug shortages belong to three primary sources: price of products, quality issues, and manufacturing capacity constraints (6,9–11). Additionally, the COVID-19 pandemic caused lockdowns in China, India, and Europe, unveiling the heavy dependence of the US supply of critical medicines on products manufactured in these countries (6,12). The shortage of drug supply due to the pandemic is actively tracked by FDA and kept current on the agency’s web page (13). The shortages affect almost every type of product and medical need, from over-the-counter analgesics (e.g., ibuprofen or acetaminophen) to antibiotics, sedatives, respiratory support products (e.g., albuterol), and many others. In the case of APIs, the magnitude of the US dependence on offshore manufacturing is enormous; at least 72% of drugs used in the US are manufactured abroad (6). The impact of this trade imbalance on the US trade deficit is tens of billions of dollars per year. The need for drug manufacturing in-source balancing is immense. The regulators are aware of the need to increase the domestic manufacturing capacity and our ability to reshore them (6,12,14).

The fundamental question is: Are we ready to execute a massive initiative focusing on API and dosage forms reshoring? Considering the cost elements that affect current generic manufacturer decisions, including the labor intensity, the environmental controls, and capacity constraints, the answer is that we have gaps in the technology available to support this endeavor successfully. The follow-up question is: Are we capable of such transformation? The answer is likely affirmative, in particular if we take full advantage of advanced manufacturing methods that can enable many advantages, including a substantial reduction in the cost of manufacturing in general and the cost of labor in particular. Let’s consider the required steps.

Continuous manufacturing (CM) and process analytical technologies (PAT) available for final dosage manufacturing can be used to expand domestic manufacturing capacity rapidly, at moderate cost, and at significantly lower operating cost than the batch methods that are prevalent in overseas manufacturing activities. Companies can build CM lines capable of real-time quality control, with the capacity to make more than one billion tablets (or capsules) per year in a single room using dry blending/direct compression continuous manufacturing (DCCM) methods. Existing integrated PAT tools can enable assurance of quality of 100% of the product stream, to allow real time release of products. Moreover, in such a situation where quality can be assured for essentially every unit of product, processes can be developed and should be approved by regulators much more quickly.

Mechanical and computer systems integration for such advanced manufacturing systems is possible within a few months, in a small footprint (i.e., 1000-ft2 is enough), and at an affordable cost (in the order of five million USD investment per line). Building such lines on movable skids to make them portable and capable of manufacture at distant geographical locations is also feasible, as demonstrated by projects such as the collaboration between Pfizer, GSK, GEA, and G-CON (15).

Emerging methods for pre-conditioning APIs to make them more manufacturable are radically expanding the range of applications of continuous direct compression. In fact, as demonstrated by recent work at Rutgers University in collaboration with Janssen Pharmaceuticals, products can be developed using DCCM methods in a matter of a few weeks of effort per product due to the knowledge-based available leverage, including material property databases, libraries of models for process equipment, rapidly growing capabilities in PAT, demonstrated soft-sensing methods for predicting product quality attributes (including dissolution profiles), and growing expertise in application of automated process control.

The primary constraint for reshoring generic drug manufacturing pertains to the API. 

 shows the nature of the limitations that need to be considered and the strategies to mitigate them. Amongst the most important is the availability of locally available intermediate materials used in the API synthesis. Moving the final stages of API manufacturing to domestic facilities might well be an exercise in futility if it does not include a strategy to obtain the intermediate materials from local sources. Otherwise, we would simply be moving our dependence on external sources upstream on the supply chain, while leaving our supply chain vulnerabilities intact.

Design and integration of unit operations for continuous flow synthesis and purification started several years ago. Still, aspects related to scale-up, flexible design of the synthesis platform, and optimization through flow-sheet modeling needs additional development. Interestingly, since the basic process science for liquid-based systems is significantly better developed than for powder-based processes, very rapid gains are possible in integrating advanced manufacturing systems for APIs, oral liquids, and sterile injectable liquids. However, FDA had approved only one API for continuous manufacturing (16), and there is no continuously manufactured parenteral product yet developed and approved by FDA. Clearly, additional experience in regulatory evaluation of such applications of advanced manufacturing is needed to enable industry to move forward with confidence.

One of the most promising new continuous flow synthesis systems is the SRI AutoSyn multistep flow synthesis system. SRI developed this continuous flow synthesis system as a multistep universal synthesizer to develop new drugs using SRI’s artificial intelligence proprietary tools. The speed at which this system can generate a small molecule synthesis route allows for the development of new synthetic pathways for generic drugs based on intermediate synthesis materials available in the US. The main technology drawback preventing its current use as a pure continuous flow synthesis is the units’ current scale, designed to produce grams per hour. Nevertheless, translation of continuous flow synthesis pathway into the traditional batch process or hybrid continuous and batch process design is an option that can accelerate the development of the needed commercial scales of high demand generic products.

An important constraint faced by companies that seek to implement advanced manufacturing methods, in particular generic-drug companies with limited experience in this area, is that advanced manufacturing is knowledge intensive. Many decision makers at generic-drug companies currently have limited visibility into the entire set of issues required to implement these technologies successfully and to develop processes that can be approved by regulatory agencies. In the past 15 years, the authors of this article have supported many such efforts to successful conclusion. In our opinion, access to knowledge and expertise, and management commitment, are by far the most significant elements required for success. Fortunately, due to the efforts of C-SOPS based at Rutgers University, the Massachusetts Institute of Technology, the National Science Foundation, FDA, and leading US-based pharmaceutical companies, such as Janssen, Vertex, Pfizer, and Eli Lilly, the US is in a position of leadership in access to the knowledge base required to implement advanced pharmaceutical manufacturing applications.

Securing medicinal product availability in the event of a supply-chain disruption is a problem that needs a robust solution. An effective approach will take years to implement, which simply means that we need to start as soon as possible, and we need to mobilize resources based on a lucid, long-term implementation plan. Patients worldwide, particularly our fellow US citizens, need to have affordable, available, and high-quality medicines produced within our geographical boundaries, to ensure that they are accessible under any supply-chain circumstances. The advances in pharmaceutical technologies achieved in the US are crucial elements of an innovation strategy that promises to solve the issue sustainably. It is a matter of starting and getting the work done as soon as possible.

C.C. Markides and N. Berg, “Manufacturing Offshore is Bad Business,” 

East West Manufacturing, “What Are the Benefits of Offshoring?” 

P. Van den Bossche, et al. (2020). “Trade War Spurs Sharp Reversal in 2019 Reshoring Index, Foreshadowing COVID-19 Test of Supply Chain Resilience,” 

 online, DOI:10.1016/j.ijpe.2018.08.027 (Nov. 2018).

FDA, “Safeguarding Pharmaceutical Supply Chains in a Global Economy. Testimony before House Committee on Energy and Commerce, Subcommittee on Health,” 

Robotics and Computer-Integrated Manufacturing

D. Gorenstein, “Why Pharma Companies Are Bowing Out of Generics,” 

GAO, Generic Drugs Under Medicare: Part D Drug Prices Declined Overall, But Some Had Extraordinary Price Increases (August 2016).

C. Rowland, “Investigation of Generic ‘Cartel’ Expands to 300 Drugs,” 

J. Saumya and B. Aakash, B. (2019). “Mylan Gives No Clear View on Strategic Options, Shares Plunge 17 Percent,” 

K. Sheikh, “Essential Drug Supplies for Virus Patients Are Running Low,” 

FDA, “Drug Shortages: Root Causes and Potential Solutions,” 

S. Gottlieb and J. Woodcock, “FDA Is Advancing New Efforts to Address Drug Shortages,” 

Pfizer, “Pfizer Announces Collaboration with GSK on Next-Generation Design of Portable, Continuous, Miniature, and Modular (PCMM) Oral Solid Dose Development and Manufacturing Units,” Press Release, Oct. 29, 2015.

S. Lee. “Continuous Manufacturing with a Generic Perspective,” presentation at Regulatory Education for Industry (REdI): Generic Drug Forum (Online, April 3-4, 2019).

Eric Sanchez is a senior consultant of Integra Continuous Manufacturing Systems. Fernando J Muzzio, PhD is the President of Integra Continuous Manufacturing Systems; distinguished professor of the School of Engineering at Rutgers, The State University of New Jersey; and member of the Pharmaceutical Technology Editorial Advisory Board.


Vol. 45, No. 2
February 2021
Pages: 28–30

When citing this article, please refer to it as E. Sanchez and F. Muzzio, " Reshoring Pharmaceutical Manufacturing to the US: Can We Do It?," 

Articles

Advanced manufacturing technologies are available, but challenges need to be addressed.

 Reshoring Pharmaceutical Manufacturing to the US: Can We Do It?

The pharmaceutical industry handles large volumes of granular materials such as powder blends for tablet production and filled capsules everyday (1, 2). Slight changes in ingredient properties or process operation conditions can have a major effect on a finished product's quality. Given the market and regulatory uncertainties that are commonly associated with drug product development, pharmaceutical companies typically have several drugs in various developmental stages at the same time. Because of this volume, the industry must have computer-based rapid-prototyping tools that can efficiently capture and resolve the technical aspects of drug product development so that companies can confidently make decisions about drug portfolio management and planning (3, 4). 

Rapid prototyping offers several potential benefits for design exploration and manufacturing process validation. Methods such as computer-aided equipment design, computer-aided manufacturing systems, mechatronics, enterprise software, and computer-graphics–based imaging have become widespread in health-related fields (5–7) such as device design and high-end surgery. 

Partly as a result of advances in the process analytical technology movement, computer-aided modeling also can be used to track the motion of granular entities that govern the flow and mixing of the solid ingredients in drug manufacturing. Although practitioners have attempted to handle powder-flow problems using empirical approaches in the past, it is clear that enhanced predictive power—rooted in more detailed studies of powder-flow physics—is required to develop strategies for the systematic design, optimization, and control of pharmaceutical manufacturing processes. Over past 10 years, considerable work has been done in this direction. For example, Barker (8) and Bertrand (9) have conducted comprehensive surveys about computer simulations of the dynamic behavior of granular material, structure, flow, and mixing.  

This article describes current, state-of-the-art modeling techniques and discrete element methods that are increasingly used to model and simulate pharmaceutical granular flows. We describe obtainable simulation results for free-flowing and cohesive granular material using rotary blenders as an example and emphasize current limitations and challenges that must be explored in future work. 

Granular-flow models can be divided into three broad categories: continuum, kinetic theory, and discrete. 

 neglect the discrete nature of grains and assume a continuous variation of matter that obeys mass and momentum conservation laws (10–11). Thus, the material's behavior is described by constitutive equations, frequently 

, that relate kinematic, mechanical, and thermal field variables. Some continuum models incorporate microstructural parameters that describe characteristics and mechanisms that are distinctive of granular materials such as the solid fraction distribution and dilatancy phenomena (12).  

 exploit similarities between interacting grains and colliding molecules in a dense gas (13). They incorporate energy dissipation and take into account fluctuational velocities with a Maxwellian distribution. They solve continuum hydrodynamic forms of mass, momentum, and kinetic energy conservation laws (14).  

 have numerous subclassifications, but they all assume that the constituent grains are distinct and move according to prescribed rules. Examples of discrete models include Monte Carlo methods (15), which apply probabilistic rules; cellular automata (16), which use deterministic (and often simplified) rules; and particle dynamics, for which rules are derived from momentum conservation. Two simulation methods typically used for granular materials are based on hard- or soft-particle approaches.  

 assume that particles interact by instantaneous binary collisions and collisional operators to dissipate energy. They are used for diluted systems and thus are more appropriate for rapid granular flow (17).  

, collisions are not instantaneous and last for a finite amount of time (

enduring collisions). These models are more appropriate for dense granular flows (

 volume fractions larger than, or equal to 50%), thereby allowing the handling of multiple enduring contacts (18–20). In this situation, particles are allowed to have deformations, which are used to compute restoring elastic, plastic, and frictional forces.  

When two particles collide, they usually undergo a deformation that may fall somewhere between the extremes of perfectly elastic or perfectly inelastic states. In the soft-particle model, collisions are inelastic, but the deformation is an overlap distance— the larger the value, the larger the repulsive force, and as a consequence, the larger the kinetic energy loss. In this process, the particles may also slide relative to one another, resulting in frictional forces. These considerations have been taken into account in the soft-particle method used in the study described in this article. The method also can be used to simulate chute flow (21), heap formation (22), hopper discharge (23), and flows in rotating drums (24).  

In the present study, the granular material was a collection of frictional and inelastic spherical particles. Each particle may interact with its neighbors or with the tumbler walls under normal and tangential forces. The normal forces that develop between particles-in-contact are calculated using the Walton and Braun model's "partially latching spring" (25). Tangential forces are calculated using Walton's three-dimensional "incrementally slipping" model (26). 

Previous work investigating cohesive flow involved the addition of liquid to granular systems (27). Nonetheless, the dynamics of pharmaceutical materials with dry cohesion are substantially different from those of wet cohesive materials. More recently, the cohesive force between particles was simulated using a square-well potential. To compare simulations with various numbers and particle sizes, the force's magnitude was represented in terms of the dimensionless parameter,  

 is the bond number and is the measure of cohesiveness that is independent of particle size, 

 is the cohesive force between particles, and 

 is each particles' weight. This constant force may represent short-range van der Waals or electrostatic effects. 

Granular materials flow and mixing simulations

The soft-particle dynamics algorithm has been used to report the first three-dimensional simulation of a tumbling (double-cone) blender for free-flowing particles (28). Previous studies of mixing and segregation in the double-cone blender provided the experimental data needed to validate the models (29). According to these studies, blenders with reflectional planes of symmetry orthogonal to the rotation axis suffer impeded transport across the symmetry plane. A synopsis of early results appears in Figure 1. 

 Figure 1: (Top) Front view of a time sequence from a simulation of monodispersed particles differing in color in a double-cone blender. (Bottom) Corresponding experimental snapshots for a transparent blender with identical operational conditions. N denotes the number of revolutions.			

In this article, we followed 15,000 identical 2-mm diameter spheres (44% of total fill volume) through 6 blender revolutions at 15 rpm. Particles were loaded first from above in random, nonoverlapping positions and were allowed to fall to rest before the simulation began. The simulations (Figure 1, top row) showed the essential convective transfer (Figure 1, bottom row). After 6 revolutions, the yellow particles appeared uniformly dispersed with the right half of the tumbler, but there was very little convective transport between the two halves of the tumbler. In both experiments and the simulation, particle motion consisted of an azimuthal convective flow accompanied by distinct axial diffusive transport. 

Figure 2(a) shows the prediction of segregation patterns and displays a top view of a simulation using a bidisperse mixture of approximately equal masses of large and small beads. First, 11,500 small (6-mm diameter, blue) spheres were mixed randomly with 1500 large (12-mm diameter, red) spheres. The blend was rotated at 30 rpm for 12 revolutions. For comparison, a snapshot was taken of an experiment in which an equal mass blend of 1.6-mm (blue) and 4-mm (red) diameter glass beads were rotated at 16 rpm in a 1-qt double-cone blender (see Figure 2[b]). In both the experiment and the simulation, the tumbler was filled initially to 50% of its total volume. 

Figure 2: (a) Top view of the segregation of 12-mm (red) and 6-mm (blue) spheres in a double-cone tumbler after 12 revolutions at 30 rpm. (b) Top view for comparison, using 4-mm (red) and 1.6-mm (blue) glass spheres. (c) Slice through the vertical plane passing through the blenderÃÂ´s axis of rotation, showing interior structure in the simulation. (d) Slice through the corresponding vertical plane in the experiment. 			

The same segregation pattern occurred in the tumbler's first few rotations: Differently sized particles predominated in a shell and core pattern along the blender's axis of rotation. In the experiment, the pattern changed little after the first few revolutions. Figure 2(c–d) shows the segregation patterns in the tumbler's interior for the simulation and experiment respectively. These figures show a sideview of the blender intersecting a vertical plane through the blender's rotation axis (indicated by green arrows in Figure 2[a, b]). The simulations and experiments showed that the small grains formed a contiguous core parallel to the rotation axis.  

To produce the experimental view (see Figure 2[d]), a thin, metal plate was inserted along the vertical plane of interest to divide the blender into equal halves. Melted paraffin wax was poured on the top half of the blender. Then, the grains were vacuumed away from the other half (see Figure 2[d]). Once the paraffin settled, it held the grains in place so that the tumbler could be tipped with the metal plate facing up. Then, the plate was removed, and the topmost exposed layer was vacuumed to reveal the interior structure. 

The previously used discrete element model was expanded later to include cohesive and frictional forces to characterize the behavior of cohesive systems. The model system used in the remaining examples shown here consisted of 20,000 2-mm diameter particles in a cylindrical vessel with a 9-cm diameter, 1-cm length, and frictionless end caps. Although the number of particles was much smaller than that used in experiments, the intensity of cohesive forces per unit volume was matched by direct observation of flow features. The model captured the main experimental observations.  

In other experiments, the cylinder was loaded with the powder to 50% by volume and rotated at 7 rpm. The model displayed discrete, well-defined avalanches (see Figure 3) for the moderate cohesion scenario, with a 

 value of 60. The flow of particles in the simulation compares well with the flow of Avicel 102. The largest avalanches were at the onset of flow, settling to a smaller, nearly constant avalanche magnitude as the bed dilated. The volume-averaged cohesive forces decreased because of a smaller density of interparticle contacts. 

Figure 3: Typical avalanching behavior, captured in a time sequence, displayed by the DEM model (20 rpm) for K = 60. The bottom row shows experimental snapshots (7 rpm) for A102 showing essentially identical behavior for comparison. 			

To test the model further, the material's dilation was measured in the simulation and compared with that of the experimental results. A custom-designed pixel-counting program was used to determine the relative volumes of powder to void. The bed dilation was uniform along the cylinder axis; visual observations support this assumption. Pixel data were analyzed for snapshots taken every 0.3 s for simulations and every 1 s for experiments. The bed's initial volume at time (

Dilation was quantified as the percent increase in bed volume relative to the initial volume. The last frames were taken at 

 = 0 before the onset of powder movement. 

The dilation of a few other powders of differing cohesion were measured and compared with simulations of various 

 values. In Figure 4, the three powders of increasing cohesion yielded dilation values of approximately 10, 15, and 24; with the simulations at 

= 45, 60, and 90, dilation results of 10, 14, and 20% were obtained, respectively. Both the simulations and the experiments show the same trend of increasing dilation with increasing cohesion. The slight differences in the magnitude of dilation may originate from the very different vessel–particle size ratios used in the experiments and simulations. 

Figure 4: Effect of cohesion on dilation. Experiment (top) performed using powdered lactose, Avicel-102, and Fast-Flo Lactose. Simulation (bottom) corresponds to K = 45, 60, and 90.			

The effect of cohesion on the mixing and segregation of the uniform binary mixture also can be captured by simulations. The computational model system for this case study consisted of 10,000 green and 10,000 red identical particles of 2-mm diameter, loaded initially side-by-side along the axis. The mixing drum in the simulation had a 9-cm diameter and a 1-cm length; It was rotated at 20 rpm. Four cohesion values were simulated with our model system. Snapshots for each value of 

 were compared (see Figure 5). Snapshots were taken at 

Figure 5: This figure shows the effect of cohesion on mixing.  The mixing of highly cohesive materials (K = 90) is slower than for cohesion-free materials (K = 0). For mildly cohesive materials (K = 0.1), the mixing is faster than for the free-flowing materials (K = 0).			

 = 0) blended faster than the moderately cohesive (

 = 90) cases, for which red particles moved in unison, forming a stretchable band. The very mild cohesive case (

= 0.1) showed faster mixing than the free-flowing case, which is in agreement with previous experiments (30). 

Particle-dynamics simulations were applied to characterize the flow and mixing in partially filled industrial tumblers with uniform free-flowing and cohesive binary mixtures. Our results show that dry cohesive powders dilate substantially and exhibit avalanches. Their size is governed by the intensity of the cohesive forces. The discrete element model captures the main phenomena observed in experiments. Mixing slows down for highly filled mixers. Axial mixing is much slower than radial and angular mixing. Particles of different sizes segregate in a shell-and-core pattern. Low-intensity cohesion enhances mixing. Despite these results, high cohesion values (typical of commonly used pharmaceutical powders) show slower mixing, which is attributed to the formation of plug flow in the cascading layer, causing sluggish thinning of striations of similar particles.  

Although the examples used in this article focus on blenders, discrete element models also can be used to simulate flow, mixing, and segregation for other applications. By appropriately changing the boundaries and tuning of friction-cohesion forces, the model can simulate die filling, compaction and decompaction of the cohesive granular systems in tablet-manufacturing processes, hopper flow, wet and dry milling, agitated dryers, motion through transitions and feeders, and numerous other applications, leading to an enhanced understanding of the effects of particle properties, system geometry, and operating conditions on the outcome of manufacturing processes. The field is advancing rapidly and substantial progress in anticipated in the next few years.  

 is a professor in the department of chemical and biochemical engineering at Rutgers University, Piscataway, NJ 08855, tel. 732.445.2972, fax 732.445.2581, 

 Dr. Muzzio also is a member of Pharmaceutical Technology's editorial advisory board. 

*To whom all correspondence should be addressed. 

Pharmaceutics the Science of Dosage Form Design

 (Churchill Livingstone, Spain, 2002, 2d ed.). 

Pharmaceutical Powder Compaction Technology, Drugs and Pharmaceutical Sciences

 (New York, NY, Marcel Dekker Inc., 1996). 

 "Biopharmaceutical Process Development, Part III: A Framework to Assist Decision Making," 

 "A Software Took to Assist Business Process Decision Making in the Biopharmaceutical Industry," 

5. G.G.K. Jacob, C.C. Kai, and T. Mei, "Development of a New Rapid Prototyping Interface," 

, "Rapid Prototyping of Real Time Control Laws for Complex Mechatronic System: A Case Study," 

Computer Aided 3-D Simulation and Prediction of Craniofacial Surgery: A New Approach," 

8. G.C. Barker, "Computer Simulations of Granular Materials" in 

Granular Matter: An Interdisciplinary Approach

, A. Mehta, Ed.(Springer-Verlag, New York, NY, 1994), p. 35. 

9. F. Bretrand, L.A. Leclaire, and G. Levecque, "DEM-Based Models for the Mixing of Granular Materials," 

10. M.A. Goodman and S.C. Cowin, "A Continuum Theory for Granular Materials," 

The pharmaceutical industry handles large volumes of granular materials such as powder blends for tablet production and filled capsules everyday (1, 2). Slight changes in ingredient properties or process operation conditions can have a major effect on a finished product's quality. Given the market and regulatory uncertainties that are commonly associated with drug product development, pharmaceutical companies typically have several drugs in various developmental stages at the same time. Because of this volume, the industry must have computer-based rapid-prototyping tools that can efficiently capture and resolve the technical aspects of drug product development so that companies can confidently make decisions about drug portfolio management and planning (3, 4).