Anomalies can pose a significant impact on safety and production goals, particularly in the chemical processing industry. When these disruptions are not identified quickly, it hampers root cause analysis and, thus, lingering process upsets.
There is no question that chemical engineers are experts in understanding their production processes. But when anomalies arise, they may face challenges in determining root causes without engaging data scientists to provide the analysis. Oftentimes, however, data scientists are not readily available. Even if they are, tutorials may be required on specific processes in order to provide the data needed for root cause determination.
But today, with advanced process analytics, chemical engineers are finding ways to run analyses more quickly and without the need for data science techniques. Self-service data analytics has empowered chemical engineers to become data scientists, greatly improving production processes. This self-service approach has delivered far-reaching benefits including greater sustainability, improved operations performance, plant safety and increased value from the communications chain.
Unlocking data analysis for chemical engineers
Clariant, a multinational specialty chemicals company, has a dedicated department of data scientists. However, the company is moving from a classic data analysis approach to a more hybrid organization where everyone in the data science department is also a chemist or chemical engineer.
In its mission to create this hybrid model, Clariant sought the expertise of a partner that could successfully help guide its digitization initiatives. It chose an organization that could deliver specific expertise with a self-service software solution that could empower chemical engineers to analyze time-series data without the assistance of a data scientist.
Initially, Clariant deployed the data analytics solution at its German plants. With proven success, Clariant rolled out the software solution to multiple sites with self-service analytics now being utilized at Clariant’s production sites around the globe.
The company’s specific objectives for the initiative included breaking down communications silos at individual sites in order to compare process behaviors across the entire organization, tasks that were not previously possible.
Additionally, Clariant’s investment needed to accomplish two key business goals: to decrease the amount of raw materials used globally in production processes and to reduce cycle times of batch processes.
Analytics phases for digital maturity
Clariant’s digital maturity has evolved and advanced to result in a global plant structure that is highly efficient and data-driven.
Before Clariant began its advanced analytics journey, the company clearly defined five phases of development to meet its mission of reaching digital maturity.
The first phase is a descriptive analysis to determine what has happened. A diagnostic assessment follows in phase two to determine why an issue occurred. Phase three is a predictive approach to determine what might happen next. A prescriptive review is conducted in phase four to identify what actions might enable the issue to recure. Finally, phase five is a cognitive analysis to identify best practices. While the first two phases incorporate monitoring and reporting, the final phases encompass advanced, or cognitive analytics.
Clariant has currently reached the predictive phase and is beginning to determine what steps it will need to take to reach the prescriptive phase, and, finally, the cognitive phase.
Clariant first established a cloud-based data lakes model that takes information for manufacturing execution systems and edge devices, as well as systems for laboratory information and production management. Raw data from these sources is then delivered to:
- A historian, to which its data analytics platform is connected.
- An organized and trusted database that filters to a sandbox and eventually is accessible company-wide.
To intensify its data analytics solution, Clariant needed to identify individual site needs and establish a user community. This involved providing learning packages in conjunction with its software partner and from its own use cases, tracking and realizing the benefits of the solution, and maintaining engagement and momentum at all sites.
Each site receives a custom training package to address specific needs, such as analysis, adoption and rollout, site-specific coaching, tracking use cases and building a community to share experiences. The sites also have a specific person assigned as the team leader to provide support and customization for that particular site.
Machine learning and Python
Clariant also uses a feature recently included in its partner’s latest software release to evaluate its toughest cases: Python notebooks. The notebook feature uses a popular computer language to help chemical engineers apply machine learning (ML) models that can provide a deeper and more detailed assessment into process behavior.
Root causes for abnormal process behavior are often difficult to determine but Clariant engineers found that the software solution was able to find root causes of process anomalies 85% of the time. To close the 15% gap in identifying process issues, Clariant used the integrated Python notebook feature to apply ML capabilities.
From its historian, Clariant’s chemical engineers use the self-service solution to gather time-series data. They then apply their own algorithms and the company’s data science platform. Finally, they create analytics on top of that model, discuss the results and get opinions on the next steps to determine corrective actions.
Computer programming is not generally within the purview of chemical engineers. It usually requires additional learning outside the scope of an engineer’s training. But Clariant engineers discovered that, with the new solution, ML coding was not going to require another advanced degree.
Python has been around since the 1980s. Invented by a Dutch programmer, it was designed to create a language that was powerful but easy enough for anyone to learn and use. Data scientists use Python as it is able to sort large datasets using short snippets of code. But the language can also be used for a variety of tasks, including establishing ML techniques.
Clariant took advantage of the Python integration by creating new dashboard features in the data analytics solution. Engineers then used different visualization types by creating ML tags in Python that were not originally available in the self-service data analytics software. This enabled the engineers to supercharge the company’s digitalization program with ML capabilities that allow process experts to establish an even stronger golden fingerprint. From this fingerprint, Clariant’s chemical engineers are able to create better monitors and alarms for key stakeholders who can then intervene in time to correct an anomaly.
Greater efficiency across operations
By increasing their skill set with the addition of ML capabilities, Clariant’s chemical engineers have gained greater efficiency across the company’s operations. Python notebooks and ML techniques have proven to be powerful tools as well as exceptionally useful features in the company’s data analytics program.
The company has combined the advantages of a classic engineering model and a data-driven analytics model. While each Clariant production site has its own profile and challenges, all have been able to increase throughput to improve production and profit.
The solution has also helped to bring people and cultures together to work on global projects. Engineers are now able to see the positive results of applying advanced self-service data analytics to solve daily problems they had been unable to figure out on their own and share that knowledge across the enterprise.
With self-service analytics and the power of machine learning, Clariant is well-prepared for the next phases of its digital journey.
Daniel Mϋnchrath is a customer success manager at TrendMiner.
Nimet Sterenberg is a data scientist at Clariant.