The Future of Intelligent Process Sensors
Meeting the PAT Framework
It’s been 15 years since the FDA first released its Process Analytical Technology (PAT) initiative, challenging the pharmaceutical industry to adopt quality innovations through real-time monitoring and control of their processes. While the initiative has indeed spurred valuable new technologies, the journey to a fully quality-by-design future is still in its early stages with monitoring of many critical process parameters (CPPs) having room for progress.
Bioprocess teams aspire to prevent rejection of batches, reduce production time, and increase yield while striving towards the PAT initiative. To do so, they need to have real-time measurement of their processes and full documentation for Good Manufacturing Practice (GMP) and root-cause analysis. Full adherence to PAT also requires a paradigm shift to real-time monitoring of CPPs whenever possible.
As a sensor manufacturer, Hamilton Company works to provide tools facilitating compliance with the PAT framework. We facilitate the move to real-time monitoring with new technologies and new sensors. We have taken advantage of technical trends, such as the miniaturization of electronics, to provide smarter sensors with more robust data, more information about the sensors, the ability to store calibration, and elimination of manual documentation. Innovations such as these enable our products to be integrated with the manufacturing environment of forward-thinking plant operators and support their movement toward the smart factory of the future.
This article highlights where plant operators, technicians, and sensor manufacturers have been in this effort, where we are now, and where we are going in the future.
Where We've been
The Hurdles of Process Sensor Advancement
Conventional sensors for pH, dissolved oxygen (DO), and other CPPs are considered analog, or non-intelligent devices. They are composed of glass, metal, and plastic with very limited electronics. It is impossible to ascertain PAT-relevant information from the sensor, such as whether the sensor is still in good condition, how many times it has been used in a process, or the number of sterilization-in-place (SIP) or other cleaning cycles it has been through. What’s more, keeping these sensors in top condition requires intensive maintenance, including expensive replacement parts and the type of skilled troubleshooting for which modern factories do not wish to dedicate time or resources.
Analog sensors also have several inherent deficiencies specifically related to signal transmission. Typical analog sensors produce a weak signal in a form that is not readily usable. This weak signal is carried to a transmitter by a cable that must be kept short to reduce electrostatic interferences, such as from nearby pumps. The transmitter amplifies the sensor input and converts it to a 4-20 mA signal. The 4-20 mA circuit, introduced in the 1950s, is spanned across the range of physical values from the sensor, so it can only provide one input for limited information. Transmitters pass this signal to the Process Control System (PCS) in addition to serving as a local display of sensor measurement. Analog sensors require a one-to-one coupling with signal transmitters, which contribute substantial cost and complexity upon startup.
Due to the way regulatory bodies such as the FDA clears drugs for sale, it is extremely cumbersome to change any aspect of the manufacturing process.
Another deficiency of conventional sensors is that calibration and validation must happen at the bioreactor. In production suites, this means that the technician must bring calibration supplies, including buffers, into the cleanroom environment and calibrate the sensor there, rather than in the controlled environment of the metrology laboratory. In the case of a failed calibration, there is the substantial additional burden of gowning out and back in to bring a new sensor to the line. All device details (including calibration results) have to be written down accurately in a notebook for proper documentation.
In early steps away from analog issues, some bioreactor manufacturers introduced elementary digital capability to eliminate the effects of electrostatic noise on the sensor signal. However, the analog sensors still required connection to bulky, costly transmitters, and maintenance issues made these sensors high in operational cost. Monitoring and documentation also still required a visit to the transmitter in the cleanroom environment by a technician to read the screen. Due to the way regulatory bodies such as the FDA clears drugs for sale, it is extremely cumbersome to change any aspect of the manufacturing process. Therefore, validated biopharmaceutical manufacturing processes are often essentially stuck in time with the original way the drug was manufactured for Phase II of the drug’s clinical trial until the benefits of a new technology outweigh the hurdles to change the process. Therefore, many conventional sensors are still in use, even though they impede implementation of the PAT framework.
In 2004, Endress & Hauser took a big step towards improving this situation with the introduction of digital sensors. Digital sensors allowed calibration to be performed in the lab and stored along with information on the calibration results, the operating hours, and more in the microprocessor-equipped sensor. The whole process of gowning or suiting up and going out for at-line calibration was eliminated. A digital signal meant that the data coming from the sensor would not be influenced by pumps or other interferences of the local environment. The digital microprocessor in the sensor head allows the health of the sensor to be known without a time-consuming visit to the transmitter. GMP demands that a technician manually documents that they have used the transmitter to see all available screens. Despite the deficiencies inherent to digital sensors, their introduction was a major advance, and many of these devices are in use today.
Purchase price is typically greater for digital sensors than conventional sensors due to the incorporation of microprocessor capability. The microprocessor solved issues related to the robustness of data transmission and reduced the maintenance efforts of at-line calibration. However, startup cost and space are also high due to the required transmitter. Operational costs are high due to the additional visits to the production floor and manual documentation needed with transmitters, so although digital sensors provide many benefits, there are still factors that could be improved.
Where We Are Now
Pushing Sensor Technology to Prevent the 'Failed Batch'
The FDA’s PAT framework calls for in-line process sensors that can move a measurement from manual sampling and laboratory instruments to automated control of critical process parameters. Hamilton has reshaped the monitoring landscape and brought the PAT framework closer to reality with the introduction of intelligent Arc sensors. These sensors talk directly to the PCS without a transmitter. They not only send a compensated measurement value used to control processes, but also send a host of diagnostic data that is automatically recorded on the sensor in tandem. The data recording and transmission are designed to meet or exceed FDA and GMP regulatory guidelines.
Hamilton offers pH, oxidation-reduction potential (ORP), DO, conductivity, and viable cell density sensors with the Arc technology. This sensor intelligence allows calibration to take place in the metrology laboratory or similarly controlled environment. This ensures an optimal calibration and never having a failed sensor at the reactor again. Arc technology also allows configuration settings to be cloned from sensor to sensor reducing technician time and potential for error. Most Arc sensors have a built-in 4-20 mA protocol to integrate directly into existing infrastructure. For more modern operations there is also built-in Modbus, an open digital protocol, which conveys all sensor data to the PCS. Modbus communication allows users to fully leverage the rich data available from these technologically advanced sensors. It is also a two-way communication, so the sensor can be configured directly from the PCS without the need for an intermediary.
Communication from Arc sensors was designed to be flexible enough to fit the needs of all users, so they can also be configured and monitored with PCs and mobile devices.
Technicians can see all important information about each sensor from a distance via mobile device or PC using Bluetooth® technology. This means that any issue with the sensor can be identified immediately without a visit to the transmitter and long before the end of a run. In the case of a failed control sensor this could mean prevention of a failed batch. The intelligent Arc sensors send alarms, troubleshooting information, quality indicators, and diagnostics both to the PCS and wirelessly to mobile devices. The information gathered and stored by the Arc pH or DO sensor, for instance, includes the measured value, temperature, serial number, cleaning cycles, calibration data, and much more.
Communication from Arc sensors was designed to be flexible enough to fit the needs of all users, so they can also be configured and monitored with PCs and mobile devices.
The automatic GMP-conforming documentation maintains the history of each sensor so that operators can tell when it has been used, calibration errors or warnings, interface errors (e.g. out of span 4-20 mA signal), and hardware errors (e.g. glass impedance). The information may be used immediately, as in the case of a process deviation, or for a future decision, such as when to calibrate or replace the sensor. This information is available through ArcAir software on a mobile device (Android or iOS) or PC.
ArcAir allows for each user to be assigned specific roles for compliance with FDA Part 11 and Eudralex Annex 11 regulations and advanced security. Roles are centrally maintained in a database so that users only see sensors in process groups assigned to them. Functions can also be defined by user role. Digital calibration and validation reports can be saved for electronic signature.
Recent technological advances were leveraged toward the development of intelligent sensors. Hamilton specifically developed the Arc intelligent sensor family in order to overcome the greatest challenges in previous user processes. Costly, bulky transmitters were eliminated, documentation automated, data enriched, and calibration simplified to vastly reduce the operational cost and effort and increase security. Despite all of these advances, there are always hurdles to overcome. PAT encourages more and more data acquisition, and technology will need to continue to advance for operators to acquire, maintain, and use this influx of data.
Learn More in These Guidebooks
Biopharma PAT White Paper
To learn more about how PAT guidelines apply to the bioreactor, download the white paper.
Intelligent Sensors Brochure
To learn more about the intelligence of Arc technology, download the brochure.
Where We're Going
Beginning to Imagine the Future of Smart Sensors
Hamilton is continually working on ways to simplify PAT compliance, from introducing new measurement parameters to progressing data management. Integration with cloud computing and Internet of Things (IoT) services will be the future of further simplifying the lives of plant operators and technicians. In this type of process the PCS will draw data from sensors and simultaneously dispatch the data to the cloud. The cloud will store information for process analytics, preventive maintenance, and asset management. Rather than manually tracking maintenance activities, the cloud can be used to automatically compute when it is time for maintenance. In an industry where a batch of product can be worth millions of dollars, plant managers may use this data to decide that it’s prudent, for instance, to replace a sensor after five uses. The cloud could also be used to monitor inventory and enable simple, GMP-conforming product documentation, and ordering of a required sensor. The functionality inherent to this type of infrastructure indicates that it is the most likely way forward.
In the smart factory of the future, where so much is measured and documented, old data handling systems like those using the 4-20 mA protocol will not be able to keep up. The Ethernet protocol will be required to handle the rich data needed for modern processes. With more and more data available, a variety of new and retrospective analyses will be possible and could yield continuous improvement and new ideas for improving batch quality.
With the enhanced functionality of Ethernet and the cloud, biopharma would be an excellent candidate for an application of artificial intelligence (AI).
Synthetic sensors, also known as soft sensors, are an IoT concept which could simultaneously simplify and enhance bioprocessing. With soft sensors, the digital measurements of two or more sensors are combined to yield a third measurement parameter. A soft sensor formed by cell density, dissolved oxygen, time, and temperature, for example, might tell of nutrient consumption, normally an unmeasured factor in bioprocessing but certainly one of interest. Soft sensors would be substantially easier to utilize in real time through Ethernet communication and cloud computing.
With the enhanced functionality of Ethernet and the cloud, biopharma would be an excellent candidate for an application of artificial intelligence (AI). Sensor data would be used as the knowledge of what’s happening in the tank, and AI would be used to optimize a process or define a golden batch. AI, especially when paired with digital twin simulations, may eventually help us to overcome hurdles in continuous manufacturing in biopharma as well.
Quality-by-Design is the ultimate goal of those in the biopharmaceutical industry looking to reach a level of the smart factory of the future. There is still a lot of innovation required to reach this goal, but Hamilton intelligent Arc sensors bring you one step closer to the future of biopharmaceutical production by increasing process understanding and control. The future of biopharmaceutical processing will reveal an ongoing increase in the array of process parameters that can be measured and the corresponding accuracy. Process scientists will be able to learn more and more about their processes, requiring more sophisticated data transmission and management tools. Ethernet and cloud computing will empower process scientists to utilize smart sensors to the fullest. This powerful combination will bring biopharmaceutical processes to an unprecedented level of efficiency, yield, and quality as predicted by the PAT initiative.
Thanks for Reading.
TAKE ME BACK TO THE TABLE OF CONTENTS TAKE ME TO THE NEXT ARTICLE