How to get advanced analytics out of the lab and into your operations

0

Chemical companies are sitting on data gold mines waiting to be exploited. Yet most of them struggle to capitalize on the full potential and insights of their data. How can you make advanced analytics work for you and gain competitive advantage?

For the past couple of years, chemical companies have been using all kinds of methods to maximize the return on investment of their existing plants. But traditional techniques can only get you so far. To further improve overall production efficiency, increase throughput and maximize yield, some chemical companies have turned to advanced analytics. But often with limited success, due to complex analytics programs that don’t match existing business processes, insights that bring no value, or fancy algorithms that take ages to deploy or never even leave the lab.

Thanks to smart algorithms and the capability to quickly sift through enormous amounts of complex data, advanced analytics has the potential to give chemical installations an extra boost.

These setbacks have left some chemical companies unsure about how to proceed. Still, thanks to smart algorithms and the capability to quickly sift through enormous amounts of complex data, advanced analytics has the potential to give chemical installations an extra boost. From product development to energy optimization and customer intelligence, advanced analytics offers a broad spectrum of solutions. The key is to define the right methodology, underpinned by data-savvy people and customizable technology.

 A successful methodology

A good methodology that covers the whole analytics life cycle is the backbone of your success. Ideally, four phases will bring you from vision to value in approximately 15 weeks.

Phase 1 – Assessing the analytics potential

To generate true business value it’s important to first assess the analytics potential of your organization. This should take no longer than three weeks. Start by defining the most relevant use cases for your company. They should be proof of business value, not proof of technology. Think about the value you would like to create and focus on key value drivers. What are your main business challenges? Can you quantify them? For instance, define the financial result of solving a particular business issue.

Once the outline of your business case is clear, you need to assess whether the necessary data is available, define the success criteria based on SMART KPIs and check the feasibility of implementing the case.

Phase 2 – Diagnosing the use case

To gain momentum, it’s important to apply the “find fast, fail fast” rule by building a minimum viable product in a lab environment. In four to six weeks, you will determine if advanced analytics will create actual business value during a short-term pilot. By using digital twins (a digital copy of the actual installation) you can gain insights into the real-world benefits. If analytics doesn’t generate the expected value, or if unexpected challenges arise, move on to another use case. This way you avoid being stuck in the pilot phase forever with no results to show for it.

Very often you will have defined multiple business cases. It’s always best to develop the highest-value case first: one scoring high on business value and easy to implement and scale. Pilots that deliver value fast and that scale to a sustainable impact will inspire the organization and encourage people to get on board. This is also the time to strengthen the capabilities of key users and focus on change management.

Phase 3 – Industrializing the data model

In this stage, analytics starts to become real within a time frame of 10 to 12 weeks. It’s also where most companies fail: successfully deploying the analytics solution in the actual plant by making it part of daily operations. But without this step, there is no actual business value.

A successful deployment depends on a number of things.

  • Ensure you have a clear deployment strategy. It’s crucial that analytics is an integral part of the business decision process.
  • Make sure you have a technology infrastructure in place that is scalable and that supports the digital transformation road map, including all key users.
  • Continue to invest in capability building and create engagement by organizing know-more sessions and training on the use of the technology. Advanced analytics should become part of people’s daily routines.

Phase 4 – Scaling the data model

If phase three was a success, it’s time to start scaling the algorithm and transfer the solution to plants across the organization. This is where the true business value lies – applying your algorithm to multiple production processes and different use cases. Look at how you can:

  • Accelerate the data model across similar installations.
  • Transfer knowledge and best practices.
  • Leverage people’s analytics capabilities across sites and business units.
  • Manage advanced analytics models as assets and across their life cycles by having the right governance in place.

People and technology

A good analytics methodology should always be underpinned by people and technology. Many chemical companies don’t have the necessary in-house capabilities and resources to cover all aspects of the analytics life cycle.

Invest in training to create a digital mindset and encourage data literacy in your company so that everyone is able to work with analytics on a daily basis. It’s equally important to invest in smart technologies that help you meet your specific needs and create high-impact business value. Think in advance about how both technology and people will enable you to scale your data models, accelerate and obtain maximum business impact.

Road to success

Today, analytics is everywhere. It will continue to change client expectations, how we do business and how companies operate. Using a solid methodology is the first step towards analytics success. It should help you to fast-track business value, provide a road map and vision, and embed analytics in your daily operations. This way it will help you solve the most intractable problems, support decision making and reveal opportunities you would otherwise miss.

Special thanks to my co-authors: Joline Jammaers, Kaat Tastenhoye and Sébastien Verhelst.

Demystifying Advanced Analytics in Manufacturing is an interactive training where our industry experts help you understand and assess the potential of your AI and advanced analytics potential better. Click the link to read more.

Share

About Author

Adriaan Van Horenbeek

Consultant Analytics

Adriaan holds master degrees in electromechanical engineering (2008) and industrial management (2009) and received a PhD (2013) in mechanical engineering at the University of Leuven for research on predictive maintenance in cooperation with several industrial companies like Bekaert and Atlas Copco. He worked two years as an industry asset management consultant for Stork where he performed reliability engineering projects at for example Umicore, BASF and VPK. Today, at SAS he holds the position of pre-sales manufacturing expert and generates value through analytics within the process and manufacturing industry. His background of engineering skills, management skills and data analytics skills makes him an ideal partner to his clients to develop and embed analytics within their manufacturing processes.

Leave A Reply

Back to Top