Gaining Insight from Process Control Data

In the past, process control data were used simply to control the process. With the explosion of data collection and analysis capabilities, however, data can be harnessed to do much more. Pharmaceutical Technology spoke with Dr. Hartmut Klocker, vice-president of the Pharma Market Development Board at Siemens, about how data from process control systems can be used to optimize manufacturing processes.

Integrating dataPharmTech: What can be accomplished by more effectively using process control data?

Klocker: In the past, process control data were collected in the distributed control system archives and dedicated plant information systems, and their only purposes were to control the process and visualize the status of the process in real-time and historically. Today, however, pharma companies try to gain additional information by bringing together all different types of data from many sources including control systems, programmable logic controllers (PLCs), analyzers, building management systems, and laboratories.

The analysis of key performance indicators (KPIs) using all these data will open new insights into manufacturing processes and provide the basis for process optimization while keeping the quality within tight specifications. In addition, new technologies, such as continuous processes for tablet manufacturing or the production of personalized medicine, would not work without integrating process and analyzer data and even parameters of the scheduling and planning tools. In the future, the industry will focus on using process data to predict the end-product quality and thus the expected therapeutic effect using model predictive control.

Some data generated within PLCs, smart sensors, and complex analyzers are not collected and not used at all. Collecting these data, transferring them to a database, and applying tools to generate valuable information out of the pile of data are key challenges. A cloud solution may provide sufficient storage capacity and calculation power. The data analysis can then be performed by the customer or even outsourced to the suppliers of the automation systems or the machine manufacturers by giving them access to the relevant cloud data. An example, available today, is equipment monitoring, with the purpose of getting a better understanding of the condition of the machines and the overall production process. Data could be then used for energy management and condition-based maintenance.

PharmTech: How can data from different sources be integrated and presented to users?

Klocker: Data integration platforms are key to bringing together these data. The networking capabilities of devices are increasing steadily in line with the general industry trends of digitalization. Key technologies that allow this integration are the open standards OLE for Process Control, Unified Architecture (OPC UA) and OPC Analyzer-Device Integration (OPC UA ADI), which is for complex analyzers with large quantities of multivariate data.

Cloud technologies can provide a basis for such data platforms. Different experts can then be granted access to a subset of these data to provide data analytical services. For example, equipment experts can be shown data for predictive maintenance, and process engineers can be shown data to support continuous process improvement. In addition, the common problem of overloading plant operators with control data that are not directly relevant to monitor the process can be avoided. Data can be hidden on the operators’ screens but transferred into the cloud, where other experts can look at various issues, such as machine health and performance or the relationship between climate and process quality parameters. The traditional operator screens will change, and instead of colorful graphics of process flow diagrams, process parameters and alarm messages with more focused views will prevail. The intent is to give a fast and clear picture of what is really going on, while hiding irrelevant details. Another tool is smart-phone apps that can be used to present the information needed to deal with specific tasks.

PharmTech: What are the security concerns with storing data in the cloud? 

Klocker: Cyber-attacks are a challenge that is mentioned in every industry, especially when storing process data and accessing it broadly. To create the highest level of security, it is important to modernize and integrate the automation environment, with security built natively into the system from the field-level up. Siemens uses a ‘defense-in-depth’ concept that covers plant and network security as well as system integrity.

Using dataPharmTech: How can the collected data be used for determining KPIs, such as overall equipment effectiveness (OEE)?

Klocker: KPIs, such as OEE and others, must use the whole range of data sources from automation systems, planning tools, building management systems, and laboratory management. There are two principal methods for achieving this. The first strategy is to collect all data in one large database and apply the KPI calculations to this database. This strategy is straightforward as long as the intelligent device can be easily integrated into the plant information database using an open communication standard. In the alternative strategy, the data are not copied into a database, but accessed from all the distributed databases on request. The databases accessed might be plant information systems and historical data from control systems, recipe-control systems, enterprise resource planning data, and others. Intelligent caching of data to allow quick access using internet browsers is important. The advantage of this concept is to avoid copying of the original data, which might make the proof of data consistency (according to GMP regulations) cumbersome.

PharmTech: What are some best practices for using process data to control the manufacturing process?

Klocker: Critical quality attributes (CQAs) determine the final product quality and are the basis for predicting therapeutic effect. Today, many CQAs can be measured in-line/on-line by applying process analytical technology (PAT). Data collected from experiments using the in-line/on-line measurement of the CQAs in the different unit operations of a production chain are the basis for process data modeling. Correlation of the data provides a basis for understanding the process and determining the critical process parameter (CPP) settings to obtain the right quality. Furthermore, the application of PAT, combined with model predictive control in the process control system, allows us to adapt CPP settings dynamically and produce within specification all the time.

A good example of the need to combine data from several sources is personalized medicine, which is of increasing importance to the pharmaceutical industry. The same principles in terms of manufacturing, quality control, and release of the drug apply to personalized medicine as to conventional drugs, but instead of large batches of active ingredients, a batch size becomes the drug product for a single patient. Without tight data integration and a fully paperless production process, manufacturing costs would exceed any acceptable limit. To overcome this challenge, process automation data and data from the laboratory information system have to be shared with the information technology (IT) systems for batch recording, reviewing, and releasing. Furthermore, scheduling systems are a key part aligning all production steps and ensuring planning principles like first-in/first-out prioritization rules. Extensive integration of these data sources by IT and automation systems will allow production costs to be lowered by at least an order of magnitude.

Another case that presents some challenges is transforming a batch pharmaceutical process into a continuous process. A continuous process requires a robust process design and continual monitoring, typically using online analyzers. Multivariate analyzer data and other process parameters from simple sensors must be combined and treated by statistical engines to give the operators a clear view of the process so that they can ensure that the process always stays within the quality boundaries. In the next generation of continuous production processes, it will become important to use these quality parameters to automatically adjust the process, thus closing the loop. The core of such an advanced process-control system is a data platform handling a large quantity of multivariate and univariate data in real time and applying statistical methods like principle component analysis. The easy integration of multivariate analyzer data (e.g., spectral data) following open standards is an important requirement. An equally important feature is the real-time integration to the control system for adjusting the set-point dynamically. Advanced process control methods, especially model predictive control strategies, can add the benefit of controlling the overall quality at the lowest variance within the lower and upper limits.

Article DetailsPharmaceutical Technology
Vol. 40, No. 3
Pages: 54–55, 76

Citation:
When referring to this article, please cite it as J. Markarian, "Gaining Insight from Process Control Data," Pharmaceutical Technology 40 (3) 2016.