Industry 4.0 in Biopharmaceutical Manufacturing

Biopharmaceutical manufacturing companies have an opportunity to participate in the latest industrial revolution, which is being called Industry 4.0. In this digitalized paradigm, machines collect process and product data and communicate with other machines via the Industrial Internet of Things (IIoT).  Using artificial intelligence (AI) or machine learning, machines can even use these data to improve processes without human intervention, although how to use AI in a regulated, GMP-compliant space is yet to be resolved. In fact, in many cases, the bio/pharma industry is still transitioning from manual systems to automation (i.e., “Industry 3.0”) (1).  Next-generation technologies are already providing more data that, along with advanced analytics and modeling technologies, can help increase process understanding and be used for advanced process control and efficiency improvements. 

Process modeling and digital twins

One of the hallmarks of Industry 4.0 is “big data” collected by equipment sensors and process analytical technologies (PAT). These manufacturing data can be used to analyze what is happening in a process and why, to predict what will happen in response to given changes, and-using simulation algorithms-to optimize processes, explained Lennart Eriksson and Chris McCready of Sartorius Stedim Data Analytics (2). Eriksson calls this higher level of simulation and optimization “prescriptive data analytics.” Others call it process modeling, and some give the name “digital twin” to a process model that is a virtual copy of the actual process. Modeling has become more sophisticated, enabling process designers to run virtual experiments and gain a greater understanding of their process before actually running it. 

Digital twins are also being used to predict maintenance and downtime more accurately; plan for process improvements, such as the replacement of components; and prepare alternative plans in case of malfunctions or disturbances, all without interrupting manufacturing, explains Billy Sisk, Life Sciences industry leader at Rockwell Automation. 

Sanofi, for example, is bringing digital transformation to its biopharmaceutical manufacturing facilities. Each of the company’s digital plants will have a digital twin connected directly to the sensors and data in the physical plant. “The data flows to these digital twins, giving managers a real-time view into the plant’s operation. Simulation on the model provides the level of manufacturing modularity and future flexibility required to support personalized medicine,” Sanofi said in an October 2017 press release (3). At its Geel, Belgium biologics manufacturing facility, sensors generate more than one billion data points in every manufacturing cycle. “By analyzing this data, potential deviations can be spotted and swiftly corrected. Yield is improved with adaptive process control strategies, and downtime is reduced by optimizing maintenance activities and shifting to predictive maintenance, based on equipment monitoring, which will increase overall output,” reported the company.

Advanced process control

Smart equipment. New technologies that allow better process control are improving efficiency and reducing waste in biopharma manufacturing by automating tasks or assisting operators to improve manual task handling. GE Healthcare’s XDUO, for example, is a new “smart” mixer for buffer solutions that continuously measures pH and performs automatic titration using acid and base pumps. This solution creates tighter control and saves a significant amount of time compared to manual titration, notes the company. 

PAT. The use of PAT to measure critical quality attributes in real-time is well-established in upstream bioprocessing, but more research is needed to apply it to downstream development and manufacturing (4). Work is underway: multi-attribute methods (MAM) that use mass spectrometry as PAT, for example, are being developed in cooperation with FDA’s Emerging Technology Team (4,5).

Automation and data security

As biopharmaceutical manufacturers transition to Industry 4.0, they should consider the requirements for changes to processing equipment, such as the automation of bioprocessing skids, and increased data security. 

Automating bioprocessing skids. The moveable, skid-mounted biopharmaceutical manufacturing equipment in use today has different requirements for automation than fixed equipment. Fixed equipment typically uses a distributed control system (DCS). Stand-alone skids, however, typically use programmable logic controllers (PLCs), which work well for isolated equipment but are more complex to integrate into supervisory systems, such as a DCS or a manufacturing execution system (MES), explains Bob Lenich, Global Life Sciences director at Emerson. To solve this problem, Emerson’s new DeltaV PK Controller is a DCS controller for skids that are self-contained, like a PLC, but can be more easily plugged into a plant-floor automation system, such as Emerson’s Delta V DCS (for coordinated operations in real-time with alarming, automated closed loop control, an integrated user interface, and historical data collectionthat a DCS provides) or into a MES (for stand-alone pieces of equipment that only need limited direction on what to produce). 

“In the near future, we anticipate that there will be more of two types of skids: mobile and movable skids, with the difference being that mobile skids are moved much more frequently,” says Lenich. “What’s missing today is a tracking mechanism to identify the physical locations of mobile and movable skids. Many of our customers are interested in using [radio-frequency identification] RFID tags as a potential solution for mobile skids because they need constant tracking. In contrast, moveable skids are repositioned less frequently, and most end users would be fine with something simpler like a barcode or QR code that doesn’t require constant tracking.”

The use of skids and super-skids with single-use (i.e., disposable) components presents additional automation challenges, notes Torsten Winkler, lead of the Life Sciences Center of Excellence in EMEA for Honeywell Process Solutions. “It is important to identify single-use equipment and, more importantly, that all are connected properly. Automation systems can continuously monitor when skids are connected, but there are additional challenges with identifying the disposable parts,” says Winkler. 

Single-use facilities and skid-based equipment require a significant amount of manual set-up to connect all the equipment pieces. Producing a batch in a single-use bioreactor, for example, might require up to 900 individual connections, and any mistakes can cause the batch to fail, notes Sisk. Graphical aids can help operators visualize the instructions, and equipment recognition and verification systems can be integrated into the control platform to track and confirm equipment placement. For example, the control system could be set up to require the operator to scan a barcode on a disposable part, which would become part of the batch record.

Data security. With the increasing connectivity of systems communicating over the IIoT, data security has become even more important. Systems must use modern protocols that have adequate security. Another best practice is to segregate non-GMP business data from GMP process data using an “industrial demilitarized zone” in information technology systems, note experts. Mobile connection, such as having access to process data on mobile phones or tablets, is an issue manufacturers must consider. The technology is available, but it must be managed to maintain data security. 

“Organizations have adopted ‘safety cultures’ for years, but now they need security cultures,” says Lenich. Just as a safety culture uses various programs to make the importance of safety visible to everyone in the organization, a security culture would similarly highlight that security is everyone’s responsibility and make security best practices a regular part of day-to-day priorities, he explains.

References

1.    J. Markarian, Pharm.Tech. 42 (4) 20-25 (2018).

2.    L. Eriksson and C. McCready, BioPharm Intl. 31 (3) 18-23 (2018). 

3.    Sanofi, “The Digital Plant: From Collaborative Robots to Virtual Reality,” Press Release, Oct. 24, 2017.

4.    A. Shanley, BioPharm Intl. 31 (5) 42-45 (2018).

5.    R. Deshpande, “Progressive Technologies to Realize the Vision of Advanced Biomanufacturing,” presentation at IFPAC 2018 (North Bethesda, MD, 2018).