Hill's Pet Nutrition

Background

Hill’s Pet Nutrition, a premium pet food company, was seeking innovative techniques to deliver increased volumes of their high-quality products in more sustainable ways with less production waste. The Hill’s team believed that if they could monitor and predict quality accurately during production, they could adjust the manufacturing process in real-time to ensure quality specifications were continuously being met, reducing variability, thus increasing production output while reducing waste. They further posited that the right technology could handle many standard production situations autonomously, allowing them to leverage knowledgeable production staff more fully.

A Visionary Approach

Hill’s Pet Nutrition strongly believes in responsible growth and in using advanced technologies to drive insight-driven innovation. Most recently, Hill’s has adopted TwinThread’s Industrial Operations Platform to accelerate overall manufacturing productivity.

The Challenge

Hill’s produces a variety of high-quality pet nutrition products. Business strategy dictates frequent startups and product changeovers. Each startup and changeover challenges the Hill’s team to continue producing the new product to Hill’s exacting standards in moisture, fat, protein content and density. Historically, each such change required lab tests to confirm that the recipes specifications were being met.

This testing was done off-line, taking up to forty minutes for line operators to receive the results and make informed adjustments to line conditions. If lab results showed any deviations from Hill’s rigorous quality standards, production had to be scrapped or reground until further tests indicated quality was back in range. CPK, a measure of a production line’s ability to produce material within variation standards, was lower than desired.

The Solution

Hills teamed with TwinThread, using its AI-enabled Industrial Operations Platform to deliver an in-line testing model by which predicted quality results are used real-time before finished product quality data are available from the lab. This model also predicts and alerts operators when variable product situations are likely to occur, provides recommendations for process adjustments, and enables the production team to quickly optimize quality in real-time without waiting for lab tests to complete.

The Result

CpK improved by almost 30% resulting in goal attainment – increased production output and reduced waste. By the end of the first project, Hill’s operators accepted more than 80% of the TwinThread platform recommendations, which led to ‘autonomizing’ many quality decisions – that is the TwinThread platform responds to its own alerts and recommendations in a closed loop manner, resulting in more efficient operations overall. Hill’s subject matter experts can respond to production situations much more rapidly when standard alerts are handled autonomously. Hills is using these same recommendations to drive additional innovation projects as they roll the TwinThread platform out to all Hill’s plants as well as to other Colgate divisions.