In the biopharmaceutical industry, as in any modern high-tech production, methods and tools for numerical modeling of physical and chemical processes are increasingly used to solve a wide variety of tasks, ranging from the development of new dosage forms and methods of their production to analysis of the processes of transportation, storage and delivery of drugs .

Amgen is one of the leading biopharmaceutical companies in the world. Medicines from this company help millions of people suffering from serious illnesses. Each drug is a product of hard work of scientists, engineers and technologists. Amgen specialists use multiphysical modeling as a tool to ensure the efficiency and safety of all production steps. Due to the specifics of the problems arising in this industry, it is very often necessary to deal with the modeling of complex physicochemical processes; therefore, the availability of a wide range of mathematical models in software for numerical modeling is a key factor.

Pablo Rolandi, director of production processes at Amgen, analyzed how the company's specialists use the COMSOL Multiphysics ^® numerical simulation environment to solve their problems.

Rolandi concluded that “COMSOL is a mature platform with modern development principles that allows you to create complex models, thanks to a logical and easy-to-use graphical interface, as well as the ability to solve interdisciplinary problems of physical and chemical modeling.” Rolandi heads a group of specialists who use multiphysical modeling techniques to solve various problems at the design stage and the development of new technological processes and systems. Increasingly, these solutions are becoming convenient modeling applications that are created on the basis of a calculation model using the COMSOL Application Development Environment. Applications allow you to receive and use the results of numerical simulation to a wide range of interested employees, even those who are not experts in the field of numerical calculations and computer simulation. In a year and a half, Rolandi’s group managed to create a whole library of easy-to-use applications designed to increase the efficiency of research and accelerate the development of new technologies.

Process Optimization

One of the areas in which the Rolandi group applies the tools of numerical simulation is the optimization and elimination of “bottlenecks” in production processes. As an example of successful implementation of simulation results, one can cite the problem of intensification of the drying process. The problem arose when the company abandoned the services of a contract organization and transferred production to its own factory in Singapore.

It turned out that the performance of the filter dryer, in which moisture removal and component separation is carried out, is too low, which increases the risk of unmet demand for final products. Rolandi's group was puzzled by the construction of a computational model with which it would be possible to find bottlenecks and optimize the drying process. Since the process is multi-stage, and for the first three stages, another type of equipment was previously used, Amgen specialists did not have enough data on the characteristics of the dryers, which made it impossible to create an accurate model and determine how the drying conditions affected the performance. First of all, it was necessary to evaluate the influence of two critical parameters - the rate of evaporation and diffusion in the new dryer. The specialists had to calculate a large data set and use regression analysis to process them in order to get all the characteristics of the model. Based on the developed calculation model COMSOL, an easy-to-use application was created in which you can calculate the time required to complete a full drying cycle.

Application for calculating the drying time under different process conditions. The application provides the ability to compare the calculation results with experimental data.

This application was used by process engineers who set up equipment and production processes. The application allows users to assess the impact of changes in equipment operating conditions and take measures to improve its performance.

Ensuring high standards of sterilization

Another example of a problem successfully solved with the help of simulation is the observance of standards for sterilization of shipping containers. Drugs from the factory are transported in special containers that must be sterilized according to the strict requirements of the relevant standards, because any, even slightly bacterial contamination of the drug can lead to very unpleasant or even dangerous consequences. The standard ethylene oxide sterilization procedure was not suitable for new containers and needed to be adjusted. The Rolandi group proposed to analyze the process of diffusion of ethylene oxide in new containers using numerical modeling, which made it possible to avoid numerous expensive experiments. An application was developed in which the user could specify the area of ​​contamination, indicate solubility, diffusion constants, and calculate how the concentration of ethylene oxide in the container changes over time.

An application to simulate the sterilization process, which calculates the change in the concentration of ethylene oxide in the container over time

Technologists used the application to select the optimal concentration of the sterilizing agent for the given conditions in a container of a specific geometric shape. The implementation of the application has significantly reduced, and in some cases even eliminated the need for experiments, which in turn reduced the design phase and the costs of experiments for several months. “It turned out that creating applications for numerical modeling is much more efficient,” says Rolandi.

Not only modeling

“I think not only about numerical modeling, but also about developing and integrating the most advanced applications and methods,” says Rolandi. “I believe that we are facing a strategic task, and we have just begun to address it.” One of the problems to be solved is the consideration of the uncertainty factor (inaccuracy) of the original data. In practice, the input parameters of any task are rarely set precisely, all of them are characterized by some error. To improve the reliability and usefulness of the simulation, it is necessary to take into account these errors.

For example, the Rolandi group is working on creating a computational model of an automatic injector — a device for administering a drug in an automatic mode without the participation of a doctor. The most important parameter of the device is the delivery time of the drug. This parameter must be precisely controlled in order to strictly observe the dosages of the administered medication prescribed by the doctor. The difficulty is that the delivery time is determined by a number of parameters known with different accuracy - these are the geometric dimensions and shape of the container, the density and viscosity of the preparation, the coefficient of friction for the piston in the injector, etc. If the error of these parameters is not taken into account, it is impossible to determine the variance for the time of drug delivery, and without this characteristic it cannot be precisely controlled. As a result of the simulation, it is important to obtain the probability distribution of the result; this will allow a better analysis of the work of the entire system as a whole.

To analyze the sensitivity of the delivery time to various initial parameters, the specialists from the Rolandy group used the tools of multiphysical modeling, with the help of which they calculated the sensitivity index for each parameter of the problem. For example, they found that the viscosity of the drug and the geometric dimensions of the needle determine 90% of the dispersion for the delivery time, and the remaining parameters are the remaining 10%. This made it possible to significantly simplify the model, since only a few parameters have a significant impact on the delivery time. In turn, this knowledge facilitated the drafting of technical tasks for component suppliers and reduced the risk of errors.

An application for the simulation of an automatic injector that calculates the error of the drug delivery time

Like other computational models built in COMSOL Multiphysics ^® , the model for analyzing the drug delivery time was turned into a convenient and easy-to-use application where users can view documentation, set initial data, perform error analysis and automatically generate reports on calculating. The introduction of the application again saved time and money for research.

Deploying and Distributing Applications

Amgen uses the local version of COMSOL Server ^TM to make applications available to its employees. “We want all Amgen employees to use our applications,” says Rolandi. - I am proud that at the moment the company is actively using more than ten applications. The introduction of such technologies has become possible only through the use of the COMSOL environment. ”

COMSOL Server ^TM makes it very easy to deploy, administer and use applications both inside and outside the company via the Internet. Users can simply log in through a normal web browser and access the application library developed by the Rolandy group.

Amgen application library developed by the Rolandi group

The Rolandy group is not going to stop there and plans to integrate its applications into the technological process, for example, to automate the input of initial data, and to turn the applications themselves into the “computational core” of the company's information system.

Speech by Pablo Rolandi at the COMSOL 2017 User Conference

Interesting articles on the topic from the blog COMSOL

• Optimization of biopharmaceutical processes using numerical modeling
• Analysis of fine chemical synthesis in plate reactors

Additional Information

More examples of the use of COMSOL ^{® by} research teams from ASML, TAUW, NRC, Endress + Hauser, Sintex, Amgen, TUM, EPFL, NTS, etc. can be found in the Russian language issue of COMSOL NEWS 2018 .

And we invite everyone on November 1 to the main event for current and future COMSOL users - COMSOL Day in Moscow .