Case Study

Instrumar uses ML to improve quality & efficiency in fiber production

Text: "Case study: Instrumar uses ML to improve quality & efficiency in fiber production" - Image: black and white fiber production machinery


Synthetic fiber is everywhere. In the clothes you wear, the carpet you walk on, the car you drive.

“You’d be hard-pressed to reach your arms out without touching something that contains synthetic fiber,” says Leigh Puddester, President of Instrumar.

Instrumar is an employee-owned company based in St. John’s, Newfoundland with a long history of solving technology challenges for industrial customers.

Over the past 20 years, the company has developed proprietary electromagnetic sensors and analytics software used by some of the world’s leading manufacturers of synthetic fiber. Instrumar’s solution allows them to monitor fiber as it is being produced, in real-time, and immediately respond to quality issues as they arise.

They are helping companies all over the world significantly reduce costs, customer claims and waste by producing higher-quality fiber and catching defects that traditional sampling and lab testing approaches could never find.


Prior to their engagement with Amii, Instrumar was at the Exploring stage of Amii’s AI Adoption Spectrum. Through their work with us, they entered the Implementing stage.

Amii's AI Adoption Spectrum - Includes the Stages: Exploring, Initiating, Implementing, Operationalizing and Advancing

After experiencing success with smaller-scale ML projects, Instrumar began to envision the massive potential of ML adoption for large-scale projects, seeing numerous opportunities to help augment their capabilities for measuring and monitoring real-time quality and efficiency in the production of synthetic fiber.

“We were reaching a point where we needed to get more aggressive in our pursuit," says Puddester. "We decided to look for external expertise to help us determine how best to operationalize our ML efforts and reduce the complexity.”

With a desire to build their in-house skills and abilities, they began an engagement with Amii.

The Goal: Detect & Predict Thread Breaks

"The AIPI was very helpful ... How we would've approached it without that guidance would have been very different."

Ruth Abraham, Instrumar Software Developer

The first order of business was determining an idea to pursue. Through an AI Planning and Initiating (AIPI) session, a team from Instrumar alongside Amii experts examined the company’s current data and identified possible opportunities to use this data to make predictions using AI. They then scrutinized, prioritized and refined these ideas, eventually drilling down to a single idea to pilot that had the right conditions for the team.

“The AIPI was very helpful. It led you through how to frame a machine learning problem in terms of how it could be tackled and made us think about the parameters and what would be our best chance for success,” says Ruth Abraham, an Instrumar Software Developer. “How we would've approached it without that guidance would have been very different.”

Following the AIPI, Instrumar decided to look at using ML to detect and predict threadline breaks on their clients’ equipment.

Currently, the company is using traditional trial-and-error statistical methods to detect such issues, computing statistics on the data in real-time and adjusting their approach based on the changes observed.

“Our algorithms work very well, or we wouldn’t be where we are,” says Abraham. “But there's just so much more we could be harnessing by automating that process.”

“Right now, sometimes a plant is often unaware of exactly what’s causing issues,” adds Puddester. “But our sensors already collect massive amounts of data on what is happening with our customers' fiber production. Our hope is that we’ll be able to introduce automated and advanced techniques that will look for unseen patterns in the data, find the cause of many different types of defects and ultimately be able to predict quality problems before they even occur.”

Scoping the Problem

After choosing their ML problem, Amii led Instrumar through a Project Advisement, a process which helped orient Instrumar’s teams, systems, and resources towards implementing a ML project.

The teams worked together to create a solid understanding of the domain and data in relation to the business problem. They used this understanding to identify potential ML approaches, assess Instrumar’s data readiness, survey literature around similar problems, and develop an understanding of the risks involved.

Moving forward with this information, Instrumar was able to build a plan to clean their data and, with Amii’s guidance, launch a pilot project.

Working with an Apprentice to Build a Model

Through Amii’s Internships & Residencies program, Instrumar was able to recruit a highly qualified ML specialist as an intern to execute the pilot project. The intern worked under the supervision and guidance of an Amii ML Scientist.

“We were able to do a pilot project using real data from one of our key customers who was looking for innovative approaches to getting more usable insights into their production,” says Puddester. “From that data, we developed a very promising model. We were very pleased and excited by what came out of the modeling exercise.”

“The internship was a great experience,” adds Abraham. “Seeing someone with such applicable skills working with our data that we knew so well and being able to collaborate with them to uncover insights, was really exciting.”


"We think we’ve got some really exciting data results so far, and we intend to build on that and bring the benefits of machine learning to all of our customers."

Leigh Puddester, Instrumar President

Instrumar is now working with one of their major customers to implement the model they developed in a real production environment, using real-time data. The company has developed a deeper understanding of how to navigate an ML project, from ideation to execution and evaluation. They are now hiring full-time ML talent and have also established a strategic partnership with a local university to collaborate on additional data science work.

“We think we’ve got some really exciting data results so far, and we intend to build on that and bring the benefits of machine learning to all of our customers,” says Puddester.

As for their future ML direction? Ideally, to detect, predict and prevent more issues in quality and production – identifying all types of production problems before they happen, giving their customers the opportunity to make proactive adjustments to their processes and avoid disruptions.

Puddester concluded, “We collect a huge amount of data. And we believe machine learning will be critical to helping us leverage all of that data to improve our current algorithms and to identify and solve problems that our customers aren’t even aware of, or which have traditionally been too complicated to figure out.”

“Right now, we’re really just scratching the surface of what machine learning can do for us and our customers. We’re learning quickly and we know it is going to help our customers improve the quality of their fiber to levels that were previously unheard of.”

For more information on how Amii can help your organization begin to understand and adopt AI, drop us a line or book a virtual meeting with our team!

Latest Case Studies

Connect with the community

Get involved in Alberta's growing AI ecosystem! Speaker, sponsorship, and letter of support requests welcome.

Explore training and advanced education

Curious about study options under one of our researchers? Want more information on training opportunities?

Harness the potential of artificial intelligence

Let us know about your goals and challenges for AI adoption in your business. Our Investments & Partnerships team will be in touch shortly!