The Problem
Our Client was stuck in a cycle of frequent production delays and escalating batch rejections, primarily due to out-of-spec issues for two specific critical quality attributes (CQAs) and problematic chemical reactor processes. This situation worsened because it depended on the best-educated guesses of skilled technicians, a method that failed in the absence of precise data analysis. Despite robust root cause analysis techniques and significant resourcing, these efforts also failed due to inadequate expertise and outdated systems, leaving critical questions unanswered. The mystery deepened with competing effects of two types of process deviations—particle size distribution (PSD) and sedimentation. The complex interrelated nature of the issues resulted in a staggering 25% of batches being rejected and translating to a €6 million annual revenue loss, not to mention the logistical nightmares in production planning and resource allocation.
The Solution
In addressing our Client's pressing challenges, we conducted an exhaustive AI-powered multivariate root cause analysis that scrutinized everything from raw materials and batch records to storage conditions and production timelines. By pinpointing a broader range of deviation-relevant parameters and their optimal ranges, we laid the groundwork for a transformative solution. Leveraging the capabilities of Aizon Predict, we crafted a high-precision AI model capable of determining the optimal temperature ranges plus exact volumes of media additions required. This model was powered by a synthesis of operator inputs and the dynamic process values captured by SCADA systems.
The Big Wins
The impact was immediate and profound: our Client witnessed their most stable production campaign to date, with right-first-time batches soaring from a traditional 70% to an impressive 90%+. This achievement wasn't just about numbers; it represented a paradigm shift in production reliability and efficiency.
Further refining our approach, we developed sophisticated classification models that not only forecasted deviations but also pinpointed the elusive root causes of CQA deviations and variability. The integration of AI and machine learning models into the process was a game-changer, predicting deviations with a potential to impact COGS by up to $5 million annually.
This solution not only addressed the immediate pain points of production delays and quality inconsistencies but also opened the door to sustained improvements in operations, showcasing Aizon’s commitment to leveraging cutting-edge technology for tangible client benefits.
