In the fight against cancer, when a drug is finally approved and ready to be commercialized, every dose counts. A major pharmaceutical company with a blockbuster oncology product had to find a way to get every dose to every patient that was in desperate need of it. When it comes down to the equipment and systems used to produce the product, sustaining high quality and cost-effective product manufacturing requires equipment that is reliable, robust, and managed effectively, while responding to future business requirements……in reality, your physical assets tend to manage you.
As part of an extremely complex supply chain, one manufacturing site took the initiative to get ahead of its critical assets by applying an Internet of Things (IoT) strategy for predicting when these critical assets may be prone to unplanned downtime events. A plan was developed to source, connect, collect, and analyze the data in order to manage these assets in a more predictable manner.
The site estimated approximately 400 individual pieces of equipment classed as critical or having potential product loss and/or schedule impact. As such, these assets required some level of data collection, with various types of data sourced from differing predictive maintenance variables, such as vibration, thermography, speed, voltage, current, and power. Some equipment assemblies were not suitable for mainstream IoT vibration predictive maintenance (PdM) analytics, so the site team looked into Industry 4.0 data analytics and machine learning using motor drive data (e.g., speed, current, etc.). They quickly found that there was no defined road map to deploy all PdM methods using Industry 4.0 tools. A plan was developed to assign the Industry 4.0 PdM strategy to the right equipment, resulting in a state-of-the-art system that can mitigate or even eliminate unplanned downtime on all critical equipment.
The ultimate goal was to collect the required data onto a secured vendor cloud platform, as well as an internal data lake based on a PI historian. The final piece of the collection architecture was to develop and place an asset health dashboard on top of both these platforms to establish the analyze portion of the project and form the basis for a next generation machine learning knowledge base.
With the project initially backed by various stakeholders from the business, operations, IT, reliability, etc., project funding secured, and a progressive, forward-thinking, site-based culture in place, the probability of success seemed high. As the project moved forward, minor challenges arose with the collection of data, such as the handling of structured and nonstructured data, the handling of manual forms of data still present in certain processes, and other issues. As these obstacles were tackled and resolved, the one factor that no one anticipated was the pulling of funding from the project. Once funding was placed on an indefinite hold, the collection phase halted at the current phase.
Although funding proved to be a major limiting factor, the site was and still is able to leverage the investment made to date with the IoT base installed and is still way ahead of the game from where it initially began in terms of proactively managing its assets versus the assets managing them.
