Jan Meireleire Engineering Manager Ashland
Edwin van Dijk VP MarketingTrendMiner
Predicting operational performance is a complex task in any industry. For pharmaceutical manufacturers, the ability to keep critical assets online and provide uninterrupted operations can mean the difference between life and death.
Since the dawn of Industry 4.0, process manufacturers have learned that simply connecting equipment and collecting data has limitations. Raw process data often is impossible to interpret without the help of a data scientist. Today, pharmaceutical manufacturers are learning that self-service data analytics solutions empower process experts to analyze data themselves.
With these solutions, pharmaceutical manufacturers can learn about different variables that affect the outcome of their batch processes. These include controlling parameters that determine batch quality. Process experts can use a self-service data analytics solution to define the “golden fingerprint”—the ideal period of process behavior—and compare it against periods of suboptimal behavior. From this, they can begin to learn when process anomalies may occur in time to provide corrective action.
In addition to improving batch quality, self-service data analytics solutions also can help improve overall equipment effectiveness (OEE), reduce energy costs, and schedule maintenance at the right time. These directly contribute to improving an organization’s bottom line.
One benefit process engineers gain from a self-service solution is the ability to learn when to schedule maintenance. Predictive maintenance allows companies to reach the highest level of reliability for equipment and other resources used in process manufacturing while reducing unnecessary maintenance costs. Process engineers can accomplish this by decreasing the maintenance frequency.
Predictive maintenance helps prevent equipment failure and the resulting economic effect of reduced performance. To begin using predictive maintenance, process experts first must gain a deep understanding of their processes to build data models and assess situations that create appropriate prescriptions. Companies traditionally turned to data scientists for this insight. But data scientists are scarce and do not always understand the process they are analyzing. Self-service data analytics solutions help close the gap in knowledge and communication while providing process engineers with readily available analytics. This helps them make data-driven, prescriptive maintenance decisions.
Ashland’s Digital Transformation
In the case of specialty chemicals producer Ashland, the company was transforming its production lines while enhancing a digital transformation initiative. This presented a few headaches regarding production issues that its engineers could not seem to resolve. Specifically, Ashland wanted to achieve a stable production process for each batch of its 50 to 100 products. Its engineers struggled to determine why its processes behaved normally sometimes but not others. The company decided to apply an advanced self-service data analytics solution to search for patterns in historical data. From that, engineers were able to find the root cause of the problem. They also learned when that problem likely would occur again. This discovery set up the base for creating predictive equipment maintenance schedules that met Ashland’s business objectives while improving overall operational performance.
In the end, Ashland achieved stable production across its batches, developed a “just-in-time” approach to servicing its equipment, and increased profits.
Ashland is a $5B US-based producer of specialty chemicals with 7,000 employees worldwide. It produces polymers for high-performance and high-value use in pharmaceutical and personal care applications. Because Ashland works with sustained-release APIs (active pharmaceutical ingredients), quality is imperative. The company produces the binders, coatings, and matrix for sustained-release APIs, which allows patients to take medication on a less regimented schedule.
The specialized pharma area faces heavy regulations and strict quality controls. The manufacturing process also must adhere to Current Good Manufacturing Practices (CGMP). Ashland sought to amplify the efficacy, refine the usability, ensure integrity, and improve the profitability of its product line and applications worldwide.
Ashland’s process involves switching batches on and off regularly. Quality is still the topmost concern. Reliability was a challenge, so the company needed to know how its equipment was performing and how that performance could infl uence its customers’ applications. Ashland required a superior way to increase overall profitability by improving production yield, lowering costs, and avoiding unplanned process downtime.
When the company transformed its plant’s production line, the change sparked new challenges. Ashland shifted its objective to achieving a higher product value and a lower product throughput. The company also learned it needed greater control of its non-automated processes. Further, it needed to comply with strict regulations regarding product quality.
The company decided it had to learn more about its process data to help overcome some of these challenges. It used self-service analytics software to analyze, monitor, and predict its pharmaceutical manufacturing processes.
Reliable Plant Processes
For Ashland to succeed in its digital transformation, it first had to identify four key areas for advancement regarding people, processes, and solutions:
- Further automate the plant to allocate more resources to specialty pharmaceutical lines,
- Use proven methodologies, such as Six Sigma and the DMAIC (Define-Measure-Analyze-Improve-Control) cycle,
- Make capital available for improvement projects, and
- Equip people with the right tools, such as computer-aided engineering (CAE) and advanced analytics.
The company’s manufacturing and engineering teams must maintain steady quality. Workers need to be familiar with the equipment intimately and aware of the influence a process has on the product. One of Ashland’s challenges was access to good, usable data to resolve these issues. Fortunately, the company had volumes of data dating to the late 1970s. Still, the company struggled with turning raw data into valuable data for process insights.
Self-service industrial analytics helped Ashland understand its data in a more meaningful way. The solution allowed process engineers to make data-driven decisions, which in turn led to increased profits and better compliance. Ashland uses AspenTech’s enterprise asset management software and self-service analytics software to extract historical data from the dedicated server. The self-service solution helps engineers convert the raw data into actionable information.
Finished Product Quality & Process Fill Rate
The final process in Ashland’s run is perhaps its most important. The quality of the finished product becomes most obvious at this point. In one case, the company found stable production flow, air flow, and temperature control. However, an unknown parameter created equipment fluctuations.
Ashland needed to learn where the undesirable oscillations were and why it was experiencing an improper process fill rate. Using self-service analytics, process engineers were able to visualize different parameters by using the solution’s dashboard. This allowed them to identify the issue without needing to use spreadsheets, which allowed engineers to focus on the issue rather than manipulating data.
In the past, the company tried unsuccessfully to analyze various products simultaneously so it could identify potential factors that affect quality. Using a filtering module for specific products, Ashland engineers learned they could identify a timeframe (5 hours or so) where the production run was stable. They then used the software’s search and filter features to create the batch’s golden fingerprint. To accomplish this, they isolated the timeframe and included tags that showed how different parameters affected operation. With the ideal run established, company engineers learned when the process likely would behave the same way again.
Ashland has control room operators who watch various processes 24/7. However, they cannot see everything. Establishing the golden fingerprint also allowed process engineers to set up monitors and alerts. This enables them to schedule maintenance before failures occur, and to alert key stakeholders when anomalies happen. Process engineers then can act quickly to resolve the issue.
Temperature Spike in Distillation Column
In another scenario, Ashland was experiencing overheating in its distillation tower. The distillation column stopped when the temperature exceeded 102 degrees Fahrenheit. Engineers used the analytics software to search past batches for similar events. They then were able to create tags of the various parameters that could contribute to the problem.
Ashland process engineers were able to visualize how the parameters affected the distillation column and determined the temperature spike was the result of a measurement failure. The process control system had detected a failure and switched the controller to manual, which caused it to stop. Engineers were able to correct the problem and prevent the controller from going into manual mode. Furthermore, the on-target production of GMP products increased Ashland’s throughput from 70% to more than 95%.
Author Biographies
Jan Meireleire is an engineering manager for Ashland (www. ashland.com). He is a chemical engineer with over 17 years of experience in plant operations as well as process design and technology.
Edwin van Dijk is VP, marketing for TrendMiner (www. trendminer.com). He has over 20 years of experience in bringing software solutions for the process & power industry to market.