INVISTA Transforms Operations by Optimizing Manufacturing Outcomes on AWS

A subsidiary of Koch Industries since 2004, INVISTA brings to market the proprietary ingredients for nylon 6,6 and recognized brands including STAINMASTER, CORDURA, and ANTRON. It is one of the world’s largest integrated producers of chemical intermediates, polymers, and fibers. This includes many household and industrial products we use every day—and some we hope we won’t need, like automobile airbags.

"We take the quality of our airbag fibers extremely seriously," says Elizabeth Gonzalez, analytics leader at Koch Industries and formerly the analytics leader at INVISTA. "That's why we're excited that, in addition to the careful manual inspection we've always had, we're now able to analyze automated visual inspection data and use artificial intelligence to identify opportunities to produce even higher-yield fibers. It wouldn't be remotely possible to do this if all our data was still siloed at each plant site."

INVISTA data is no longer siloed at sites around the world because of an ambitious initiative to transform its operations by moving from business intelligence (BI) to artificial intelligence (AI). The data now resides in an Amazon Web Services (AWS) data lake partially built using AWS Lake Formation. AWS machine learning tools are a key piece in enabling INVISTA to use data to achieve its goal of creating $300 million of value through improved yields, stronger asset performance management, enhanced process control, reduced working capital, and higher throughput.

Prior to working with AWS, INVISTA was limited by on-premises data storage, pre-formatted BI reporting, and time-consuming analytical processes. "With our old solution, it took us two months the first time we tried to get just one plant site's historical data into a data scientist's hands for analysis," says Tanner Gonzalez, analytics leader at INVISTA.

After evaluating cloud providers, INVISTA chose AWS as a preferred vendor for its ability to decouple technologies, support integration with third-party tools, and construct systems and workflows to find value in huge volumes of data at facilities around the world. The company started by migrating 600 on-premises servers to AWS, including multiple manufacturing applications and the global INVISTA SAP footprint.

INVISTA used AWS Lake Formation to implement an enterprise data lake based on Amazon Simple Storage Service (Amazon S3). The architecture includes a Snowflake data warehouse that relies on AWS Glue for fully managed extract, transform, and load (ETL) services. INVISTA also used AWS Snowball devices to migrate tens of terabytes of data from disparate plants into its data lake.

The company has found substantial savings by using AWS to reduce its on-premises data storage. “Through our optimization and right-sizing efforts, migrating our data centers to AWS is saving us more than $2 million a year," says Tanner.

In addition, the company takes advantage of Amazon Redshift—and, in particular, Amazon Redshift Spectrum—to enable data analysts to execute complex queries against terabytes of data. It relies on Amazon Athena to extend self-service interactive querying to any employee with basic SQL knowledge. And for data science workflows, INVISTA uses Amazon SageMaker, a fully managed service for building, training, and deploying internally developed and third-party machine learning models.

One operational benefit of the predictive analytics capabilities enabled by AWS is a significant reduction in unscheduled plant downtime. "If our manufacturing team knows when a piece of equipment is likely to fail, they can bring it down for preventive maintenance," says Elizabeth. "Before AWS, we didn't have the data or the compute power needed for models to predict failures. The improved asset performance management results in reduced downtime, decreased equipment damage, and higher revenues."

By leveraging AWS, INVISTA also powers stronger process forecasting and inventory optimization. "When we can predict orders and other factors so we know how much product we're going to make in the next 30 days or what spare parts we'll need for repairs and maintenance, we can make sure we're storing only what we'll need," says Elizabeth. "By using AWS to pull ordering patterns and other data from our SAP ERP system, we have a closed-loop, fully automated stocking process for high-moving spare parts that has shown significant return on investment."

None of these operational benefits would be possible if INVISTA couldn't perform the robust feature engineering necessary to build effective machine learning models. "With our data lake hosted on Amazon S3 and built using AWS Lake Formation, we are able to unlock large quantities of time-series data for analysis and use it to make better business decisions," says Tanner. "Procuring sufficient on-premises storage and compute power would be cost-prohibitive."

Running on intuitive, easy-to-learn AWS services is helping INVISTA achieve its goal of cultivating companywide data science skills and a culture of curiosity and experimentation. "As we worked to build data literacy throughout the organization, it helped that we were able to rely on a common AWS literacy as well," says Elizabeth. "Because everyone was getting hands-on with the console and going through the same AWS fluency training, we were all speaking the same language on a technology level and could therefore concentrate on the data problems we were trying to solve."

Personnel with less technical backgrounds can use Amazon Athena to make valuable contributions to data science initiatives. "Traditional analytics environments typically involve a lot of work by technical experts to present a relatively static view of the data to business users," says Tanner. "Because Amazon Athena enables even non-technical users to experiment and explore, it expands the number of people who are unlocking value in the data."

AWS services have helped transform how INVISTA looks at its work and thinks about itself as a company. "A few years ago, no one at INVISTA was even talking about data science," says Elizabeth. "Now, data science on AWS is central to initiatives in strategic planning, supply chain management, and manufacturing operations."

All trademarks are the property of their respective owners.

To learn more, visit aws.amazon.com/manufacturing.