All News & Insights

Uncategorized

Operationalizing Data Analytics Across the Enterprise

Deploying enterprise solutions using advanced analytics helps companies discover transformational insights.   

Engineers working in manufacturing today face a very different set of challenges than their peers ten, twenty, or thirty years ago. In the past, engineers may have been lacking the data they needed from IoT sensors to solve a problem, but today they often have so much data that they don’t know where to look first for solutions.

This is where deploying digital tools, such as advanced analytics applications, comes into play. These tools empower organizations to connect with disparate data sources, apply industry-standard modeling techniques to perform predictive, root cause, and other asset analytics, and operationalize data insights to improve production processes.

Engineering teams must be given the tools to monitor many pieces of equipment at once, in many units, or at many plants across the globe to detect problems before they occur. This is where operationalizing advanced analytics across the enterprise becomes a crucial piece of success for many Manufacturing 4.0 (M4.0) projects.

To successfully scale a M4.0 pilot project to production, three steps are required: 1) connect to the data where it is, 2) analyze the problem using data and advanced analytics, and 3) operationalize the solution across the enterprise.

Connect to the Data Where It Is

According to Anaconda’s “The State of Data Science 2020” report1 , data scientists still spend approximately 45% of their time preparing data for analytics. Process engineers face similar challenges, exacerbated by the need to access data from many different systems.

For example, if an engineer works in a remote operations center and monitors assets across the globe at different plants, each plant may have a different data historian with different naming conventions for IoT sensors. Additionally, each of these historians may have different mechanisms for accessing data and converting it to a format for analytics.

Operationalizing advanced analytics across the enterprise becomes a crucial piece of success for many Manufacturing 4.0 (M4.0) projects.  

 

With the right advanced analytics application, engineers can access many disparate data sources. All data remains in its original system of record, and it is indexed so the application can grab any necessary data on demand from any data source. These sources can include the cloud, manufacturing data from historians, and data from many other systems.

When choosing an advanced analytics application, teams should consider two crucial factors: the application’s architecture must be ready to meet your data where it is (i.e., whether it is on premise, in the cloud, or both), and it must be able to grow as the underlying data structures for your organization change (i.e., the scalability of your platform to handle more data as your company grows and/or moves to the cloud). For most manufacturing organizations, data will continue to move to the cloud at a rapid pace for the foreseeable future. Therefore, the underlying advanced analytics architecture must be able to seamlessly transition, for example, from an on-premises data historian to a cloud-based one, without impacting end user work. This capability means organizations can apply advanced analytics immediately instead of waiting until all their data is in the cloud to start gaining insights and value from an advanced analytics application.

Additional recommendations to consider when choosing an advanced analytics application for deploying an M4.0 project include security and audit trail capabilities. The application must have measures in place to protect data, such as an SOC II certification. For certain industries—such a pharmaceuticals, biologics, and food & beverage—choosing an application that has audit trail capabilities is vital because it ensures data integrity and GMP compliance.

Analyze Data Using Advanced Analytics

Once teams are is connected to data and can access it easily, the next step is analysis. The right digital tool for the M4.0 journey will accelerate the time it takes to gain insights. The goal is to enable teams to rapidly iterate on troubleshooting plant issues by allowing them to test hypotheses at the speed of thought. Regardless of industry, the types of analytics teams are trying to complete often involve improving the same outcomes: productivity, reliability, and sustainability.

Productivity:  Ensuring production outcomes are optimized is critical to improve profit margins, reduce giveaway, and ensure product qualities meet the appropriate specifications. With the right M4.0 analytics application, techniques such as continuous process verification, batch cycle time analytics, clean-in-place optimization, monitoring batch quality from reference profiles, digital twin product quality prediction, and many more can be applied.

Data scientists still spend approximately 45% of their time preparing data for analytics

 

Continuous process verification, for instance, allows teams to create rules for critical process parameters (CPPs) and statistical quality control limits, each segregated by product grade to identify golden batches. These golden batches are then used to find statistical averages over key periods of interest, and to identify deviations from monitoring boundaries. This methodology, which was once manual and time-consuming, can now be automated with the right tool, with continuous updates of recent results and deviations. This allows operations to make real-time process optimizations to ensure CPPs stay within the recommended monitoring boundaries, ensuring product quality.

Reliability: Common reliability goals include deploying predictive analytics across a fleet to increase uptime, reducing maintenance cost by proactively identifying potential failures, and improving overall availability by maximizing overall equipment effectiveness (OEE).

Predictive analytics can be applied to any type of equipment, whether it be a compressor experiencing a bearing failure, a pump experiencing a seal failure, or a heat exchanger with a coefficient that is degrading at a faster rate than expected. The process for deploying these analytics is fundamentally the same no matter the asset being examined.

Many tools provide black box machine learning capabilities for organizations to utilize in this space. Unfortunately, choosing this type of solution will often leave teams with more questions than answers, such as multiple false positive alerts, with no ability to dig in and see what is truly happening. A best-in-class M4.0 digital solution will not only alert the team of a deviation, but it will also provide context for identifying the root cause, as well as a solution.

In terms of OEE, this allows teams to automatically identify, track, and categorize performance losses to reveal hidden trends (such as a pump that keeps failing, causing the plant to have losses) associated with different units or pieces of equipment within the plant. When these types of OEE solutions are properly deployed, they help teams find process bottlenecks and maximize production.

Sustainability: Sustainability is the problem of our generation, according to the Pew Research Center, which found 37% of Gen Zers and 33% of millenials say addressing global climate change is a top concern for them personally2. Fortunately, the M4.0 revolution is unfolding in tandem with tremendous innovation in the advanced analytics space. This allows teams to iterate at the speed of thought to complete investigations related to sustainable operations.

When operationalizing data analytics across the enterprise, sustainability should certainly be top of mind for any organization, whether they are aiming to minimize the production of greenhouse gases, reduce inefficient water use, or lower energy consumption.

These types of projects allow employees to fully unleash their creative potential and innovate on solutions, and are key to attracting top-tier talent, particularly in the younger generations because these goals are often a deciding job factor.

The most foundational use cases in this area today are related to greenhouse emissions reporting. Reporting greenhouse gas emissions is a challenge for any company due to lack of standardization and regulatory requirements. Most often, greenhouse gas reporting is done with spreadsheets that need to be manually updated, often taking several days to provide a single status report.

Digital tools provide the ability to locate, aggregate, and analyze the data necessary to provide overall emissions numbers. The ability to deploy a solution that accurately reports emissions does more than save hundreds of engineering hours, it also addresses the problem of auditing the process, as every step of the calculations and data source selection can be recorded.

Enable teams to rapidly iterate on troubleshooting plant issues by allowing them to test hypotheses at the speed of thought

 

As this area matures, the goal should be to shift teams from a compliance-focused and reactive approach to a more proactive approach by continuously monitoring parameters to detect and mitigate environmental concerns before they happen.

Operationalize findings across the enterprise

The last, and arguably most important piece of the puzzle in terms of transforming teams with an M4.0 strategy is operationalizing findings. In the past, many engineering calculations were stored on a single team member’s desktop, which provided little transparency into the methods being deployed. And when that employee would change roles, change companies, or retire, the organizational knowledge behind the insights would be lost. Another challenge was that modeling was based on static data sets and only updated with new process conditions sparingly. With the technologies available to companies today, these legacy problems can no longer be part of a solution.

Employees must be empowered to not only complete analyses with the new digital tools in their toolkit, but to document their findings, rapidly iterate on solutions, and enable knowledge transfer across teams so they can push changes into production at scale to realize transformational organizational value. Using advanced analytics software, organizations can implement knowledge capture and create easy-to-configure analytics reports that enhance collaboration. These capabilities are key to embedding this new strategy into workflows and company culture for years to come.

Conclusion

Advanced analytics applications are needed to quickly create insights from data and operationalize these insights across the organization. Moving from older tools, such as spreadsheets, to modern solutions empowers engineering teams to improve productivity, reliability, and sustainability.  M

About the author:
Morgan Bowling
is an oil & gas industry professional with experience at both independent and integrated major oil & gas companies. These experiences exposed Morgan to the pains and frustrations of wrangling time series data using outdated tools. As a Principal Analytics Engineer at Seeq, Morgan helps engineers address their unique analytics problems to drive business value. She enjoys leading corporations on their analytics journeys to complete complex data analyses and increase overall organizational effectiveness.

Footnotes:

1  https://www.anaconda.com/state-of-data-science-2020
2  https://www.pewresearch.org/science/2021/05/26/gen-z-millennials-stand-out-for-climate-change-activism-social-media-engagement-with-issue/

View More