Troubleshooting Using Predictive Maintenance

Published on: 
Pharmaceutical Technology, Pharmaceutical Technology, March 2022 Issue, Volume 46, Issue 3
Pages: 24–27

Smart manufacturing transforms management of tablet and capsule equipment and processes.

When it comes to the manufacture and packaging of tablets and capsules, the earlier businesses can prepare for managing potential issues with machinery, the better they can deal with them. In this new era of smart manufacturing, predictive maintenance, powered by Internet of Things (IoT) sensors and artificial intelligence (AI), will transform the way the pharmaceutical industry maintains and manages tools and processes. As manufacturing gets smarter, efficiency and productivity will improve.

Predictive maintenance

The methods by which machinery is maintained is changing as a result of technological advancement. Up until now, there have been two real options for the maintenance of pharmaceutical machinery: reactive and preventive. A third option is predictive maintenance.

Reactive maintenance as a primary strategy can be wasteful and costly, resulting in potential product loss and machine downtime. Leaving machines to ‘run-to-failure’ is also damaging to the workings of the machine. With any machine that is running to failure, there comes a point where it almost has to be rebuilt and the whole device reviewed from the bottom up to pinpoint or repair the root of the issue. This method has a significant cost, not just in financial terms, but also in terms of time and labor (1). In the author’s experience, dealing with a wide variety of machinery, the impact on the equipment and operations of a reactive maintenance strategy has an annual financial cost 30–50% higher than preventive maintenance (2). When critical situations arise on the machines, this cost escalates yet further.

Preventive maintenance and periodic inspections do well to offset potential issues and ensure machinery is well maintained, but still require significant downtime, often running to days out of production. Scheduled maintenance also doesn’t take into account individual factors or unique usage of the machinery. Whilst standardized timings or experience might be used to plan for maintenance, it would be impossible to know for sure exactly when issues might arise, and where.

Integrating IoT and AI systems into manufacturing and packaging machines creates a connected ecosystem of machinery, where sensors allow for the possibility of accurate predictive maintenance. Surveys have indicated that the average reduction in maintenance costs (cross-industry) resulting from the implementation of an effective predictive maintenance program could be 25–30%, elimination of breakdowns 70–75%, downtime reduction 35–45%, and production increase 20–25% (1). Data collected from sensors located in key positions on machines can be utilized to acquire specific information that is analyzed by AI systems to predict, with accuracy, when maintenance, repairs, or interventions are needed. Smart connected machinery can communicate and inform management and network operating centers exactly what it needs and before it is needed, which means that machines don’t need to be checked and stopped unnecessarily. This method both prevents machine breakdown and ensures that machinery works optimally.

Preventing machine breakdown

Preventing unplanned breakdowns offers the potential for higher productivity and less waste. In smart manufacturing, embedded feedback mechanisms can provide alerts, via a remotelyaccessible user interface, to communicate issues with a machine. This communication could take the form of vibration sensors in a rotary machine to identify high friction between bearings or lubrication issues. Health conditions in the machine, such as these, can be treated early, or with minor treatment, rather than major intervention.

With this early “bad news”, time can be less of a critical factor, and decision-making processes are improved by the data available and can be guided by aftercare experts. Thus, it is possible for throughput to remain high, with the time and type of maintenance needed being planned based on assessment of the data. Maintenance could take place via remote assistance depending on the situation, and in the form of emergency parts changing for critical situations, planned shutdowns, or incremental interventions to spread out the maintenance process.



Maintenance is not just about preventing breakdown, but also ensuring machinery is operating optimally. Without comprehensive data being collected, it is difficult and time consuming to accurately assess whether or not a machine is operating at suboptimal parameters. With the introduction of sensors to monitor key elements of machinery and operations, the functioning of the machine can be constantly observed, and based on detailed insights, changes can be made to the machine to ensure it is working to its best capabilities.

Considering what is already possible by observing patterns in datasets gleaned from standard sensors reveals an idea of the kind of impact that IoT and AI will have on process optimization and prevention of machine breakdown. To take the example of one of ACG’s fluid bed machines, associates identified the correct interval time for shaking the blower by reviewing data gathered from several sensors on the fluid bed machine. Rather than waiting for an alarm to alert them of failure on the blower, they adjusted the fluid bed shaking interval time from 4 seconds to 6seconds. As a result, process cycle time was decreased from 23 hours to 18 hours, and downtime and damages were reduced, as the blower filter needed less regular replacement. To begin with, the shaking time was set to a very low 4 seconds to make sure the filter didn't get damaged due to excessive particle accumulation. However, this is akin to preventive maintenance where a machine is maintained to a degree beyond what’s required, which results in increased process cycle time.

In this case, if the interval time had been more, then the particles would have accumulated within the filter, resulting in it becoming blocked and torn; if the interval time was less than required, it would negatively impact the process cycle time. Identification of the optimum shaking interval helped achieve a balance. With more IoT sensors, on a wider range of machines, and AI analytics, businesses will be able to identify anomalies and observe patterns earlier on, and make the proper intervention to achieve targeted outcomes.

Beyond profit

Predictive maintenance can have significant cost and productivity benefits, with uptime improvements, better overall equipment efficiency, and less product waste. However, organizations need to look beyond profit and assess how these benefits can be best utilized to make further positive changes.

Safety, health, and risk

Predicting and managing issues with machinery early will lead to a reduction of safety, health, and quality risks. A reduction in machine failures and a better working knowledge of the machine should also result in a more productive workforce. Plant-level associates can work in a safer environment with fewer concerns relating to issues with machinery arising unexpectedly. They are then able to make better informed decisions to improve productivity and reduce risk.

Additionally, all the data provided by machines and analyzed by AI, may have an impact on machine design. Designs could be focused more around health and safety, with easy access for maintenance, part changes, and cleaning. These changes are yet to be seen as a direct result of predictive maintenance, because the integration of IoT systems into the pharmaceutical manufacturing sector is still in its early stages. However, it is a natural progression from the way machine design is currently performed. Today, manufacturing teams are able to make design improvements based upon feedback, regulations, and individual data provided by associates working in the field. Whilst this feedback has its benefits, it is limited by human bias and capabilities. Associates are only able to assess manually collated statistics from individual cases and will focus on those elements that seem to be important from their work perspective, leaving potentially missed opportunities. Comprehensive, consistent metadata that is objectively analyzed by AI will enable teams to gather an in-depth holistic view of machinery, safety, and operations.


Sustainability remains a challenge. From API production, to packaging and disposal, there are many issues pharmaceutical manufacturers are grappling with in this area. However, reducing product rejections, optimizing production, and improving quality will potentially have an impact on reducing waste, energy usage, and the environmental impact of operations.

Although it’s yet too early to predict the full scope of benefits, it’s likely that this careful, data-empowered method of predictive maintenance will also have an impact on extending the life of machinery, improving the sustainability and cost-efficiency of assets (3). Çınar et al. state that predictive maintenance is “one of the most promising strategies amongst other strategies of maintenance that has the ability of achieving those (minimizing equipment failure rates, improving equipment condition, prolonging the life of the equipment, reducing maintenance costs) characteristics” (4).


There is still some resistance in the pharmaceutical sector to the introduction of smart manufacturing. The industry has been slow to embrace change. However, it is becoming harder to ignore the necessity for technology adoption. Aside from the tangible benefits, companies need to look at how these systems can improve the inner workings of their business and support the wellbeing and safety of employees, as well as the positive impact that can be shared with customers and consumers. Human capital and improving the wellbeing of the global community should be the driver of metrics, although it is important to consider the reasons for reservations regarding the integration of IoT and AI, including data security and the need for culture change.

Data security

Ensuring the security of data is crucial. Sensors should only be activated once a customer has agreed to the capability and parameters of the data being collected from their machines. As with all IoT systems, device registration, via passkey or password, is essential and mandatory. Yet, to guarantee security and data integrity, especially as smart manufacturing systems become more widespread, further measures should be introduced. Data diodes that enable unidirectional data flow is one method by which companies can secure customer facilities from interference. Unidirectional data flow ensures data is only transmitted in one direction (outbound traffic from the plant), so IoT sensors cannot be affected by outside tampering (inbound traffic into the plant). Data cannot be transmitted to, but only from, the sensors, guaranteeing machinery cannot be interfered with by outside sources.

Culture transformation

Incorporating smart manufacturing solutions and predictive maintenance into operations is not as simple as a cost investment. Changing processes from traditional to digital, or stepping up the use of technology requires complete workforce cooperation. Proper training needs to be provided and preparations must be made to bring all teams on board with any changes and to ensure they are well-equipped to manage the culture change of digitalization. Upskilling of staff and operators at plant level is of particular importance, as they will be the ones implementing and working with these systems closely on a day-to-day basis.

Embracing the future

Predictive maintenance is a significant step forward from traditional ways of monitoring and managing machinery.Smart manufacturing gives organizations the ability to gain knowledge of the intimate workings and by-the-minute status of critical elements of the machine, which enables the making of informed decisions in an often less time-critical manner. Faults and issues will always arise, but by utilizing this capability, companies are provided with the benefits of early information and informed problem solving.


  1. US Department of Energy, Operations & Maintenance Best Practices, Chapter 5, “Types of Maintenance Programs”(, August 2010).
  2. ACG Internal data.
  3. J. Manyika and M. Chui, “By 2025, Internet of Things Applications Could Have $11 Trillion Impact,” (July 22, 2015).
  4. Z.M. Çınar et al., Sustainability, 12 (8211) 2-3 (2020).

About the author

Anand Rajan is head of Sales, ACG Engineering,

Article Details

Pharmaceutical Technology
Vol. 46, No. 3
March 2022
Pages: 24–27


When referring to this article, please cite it as A.Rajan, “Troubleshooting Using Predictive Maintenance,” Pharmaceutical Technology, 46 (3) 2022.