Modeling software tools are available to perform various computer-aided engineering tasks in development and manufacturing
in various process industries, including pharmaceutical manufacturing. Models and modeling software are well established in
API-production operations, such as solvent selection, crystallization, and separation, for both batch and continuous manufacturing
processes, comments Jonathan Kadane, director of industry marketing for Pharmaceuticals/Life Sciences at AspenTech. Modeling
for solid-dosage pharmaceutical processes is less mature. Bristol-Myers Squibb's Stamato comments, "The behavior of solids
from a surface energy, flow, and compaction perspective is still difficult to characterize. Better understanding of the behavior
of solids would help to build more computationally manageable and accurate models." AspenTech's solids-modeling capabilities,
acquired through SolidSim in 2012, model specific, solids-handling unit operations (e.g., screens, dryers, and cyclones).
Although this technology is, to date, primarily targeted for chemical process industries, the early-users group includes pharmaceutical
companies, notes Kadane.
Modeling software can be used in all stages of pharmaceutical production, from R&D through to quality control and manufacturing,
notes Michael Doyle, principal scientist and marketing director, Materials Science Segment, at software-provider Accelrys.
Process designers can use software to understand excipient–API interactions or optimize mixer speeds, for example. Process-centric
software captures numeric inputs about the process (e.g., formulation data, PAT sensor data) and feeds this into predictive
models that allow process developers to evaluate "what if" questions, explains Doyle.
CFD software can be used to model common operations, such as scaling up a mixing tank, and for more complex operations, such
as three-dimensional modeling of a fluidized bed (e.g., tablet coater), explains Kristian Debus, director of Life Science
Some particle flow can be modeled with CFD alone, but to capture more detail about particle behavior (i.e., how particles
interact with each other, the surrounding walls and equipment parts, and air flow), developers use discrete element modeling
(DEM) software. "DEM tracks the interaction between every particle in a numerically efficient manner, modeling contact forces
and energy transfer due to collision and heat transfer between particles," explains Debus. DEM has been used extensively to
model mixing and coating, and there is now increased interest in using DEM for granulation and compaction, says Richard LaRoche,
PhD, vice-president of Engineering at DEM Solutions.
Another type of CFD is multiphase fluid flow (e.g., volume-of-fluid method). This method can range in complexity from nonmixing
to phase mixing, which allows modeling of suspensions, for example. Models can also be done in steady state or in real time
(i.e., transient state). As complexity increases, so does the computation time and expense. When choosing how to model a process,
companies must balance the accuracy needed with the computational expense that will incur.
In the past, computational capabilities have limited applications to laboratory and pilot scales, says DEM Solutions' LaRoche.
This work, however, can be used to help devise scale up-rules for process engineers. The company recently added the capability
of running DEM on large, high-performance computing and shared-memory systems, which will enable production-scale modeling.
Another type of software uses flow-sheet modeling to describe individual unit operations and provide mechanisms for exchanging
data between these unit operations. This method combines techniques, such as DEM, CFD, and population balance modeling, and
can be used, for example, in solid-dosage process development to predict what type of process (i.e., wet granulation, dry
granulation, or roller compaction) should be targeted or to predict how upstream API properties affect downstream tablet production
(6), adds Douglas Hausner, associate director for industrial relations and business development at C-SOPS.
An increasing use of modeling in the pharmaceutical industry is being driven both by the FDA mandate to better control the
manufacturing process using quality by design and PAT and by the imperative of reducing drug-product development costs and
speeding the development process, explains Doyle. A greater focus on efficiency, agrees Kadane, will push the industry toward
more process modeling. Modeling the process first, before running experiments, saves time and money because simulated experiments
are faster and cheaper than laboratory or pilot-plant experiments. In reality, however, model development and experimental
development are run in parallel and correlated with each other, adds Debus. When used together, the process is better understood
and error is reduced.
The ultimate goal is to use predictive models to interpret, understand, and control the pharmaceutical process, adds Doyle.
Achieving this state would improve consistency and reduce out-of-specification product.
The real power of modeling software for the pharmaceutical industry, says Debus, is in modeling an entire continuous manufacturing
process, including three-dimensional models (e.g., multi-phase fluid flow, DEM) of the unit operations and one-dimensional
models (e.g., fluid flow through a pipe) of the transfer from one unit to another. Modeling the process as a whole allows
advanced process control, in which a parameter change in one unit operation is accounted for in other parts of the process
as it affects other unit operations.
Most big pharma companies are using CFD, mostly at a basic level, and smaller pharma companies are just getting started with
CFD, comments Debus. Process industries (such as chemical), manufacturing industries (such as automotive), and industries
in which flow and particle interaction are crucial (such as aerospace and nuclear), have advanced the use of modeling. The
task at hand in the pharmaceutical industry is for all parties (i.e., software and equipment vendors, pharmaceutical producers,
and academia) to work together using and improving these tools to better understand pharmaceutical processes.
1. J. Parker, et al., Powder Technol. 235 115-120 (2013).
2. W. Chen, et al., J. Pharm. Sci. 99 (7) 3213-3225 (2010).
3. D. Fichana, et al., Pharm. Dev. and Technol. 14 (2) 173-184 (2009).
4. B. Benyahia, R. Lakerveld, and P. Barton, Ind. Eng. Chem. Res. 51 15393-15412 (2012).
5. R. Lakerveld, et al., AIChE J. online, DOI: 10.1002/aic.14107, Mar. 24, 2013.
6. M. Sen, et al., Intl J. Pharm. 445 (1-2) 29-38 (2013).
Join the discussion
How is the pharmaceutical industry using model-predictive design?
Post your comments on