With Industry 4.0 being the new manufacturing buzzword, companies are working hard to make their production facilities ready for the next industrial revolution. The core principle of Industry 4.0 is having a fully connected production plant that utilizes industrial data for process monitoring all along a manufacturing process.
With countless proven examples of successful implementations, multivariate data analysis is a natural fit for Industry 4.0 objectives. Multivariate analysis enables manufacturers to utilize their historical data to improve quality, increase yield, and reduce operating costs.
ProSensus’ breadth of experience in analyzing manufacturing data from pharmaceutical, food, and chemical manufacturers allows us to offer insightful site-wide and global data audits. These audits provide our clients with a detailed understanding of what data is currently available and what insights it may provide to improve operation. Our audit will identify areas where instrumentation or automation is lacking to optimize improvement spending for your facilities.
ProSensus generally conducts a comprehensive plant-wide data audit using the following 4-Step framework:
- Factory tour
- Establish data-audit objectives
- Multivariate data analysis
- On-site & in-hand results
Case Study – Powder Metal Components Manufacturer
Recently, ProSensus completed a plant-wide data audit for a global powder metal components manufacturer.
1. Factory Tour
Process experts walked ProSensus personnel through the entire manufacturing process outlining key process units and present data collection.
2. Establish Data-Audit Objectives
Next, on-site meetings with key process managers were arranged to further discuss current challenges and initiatives within the plant. ProSensus worked with the client to review their dataset and to appropriately format it using ProSensus’ data template. This is a critical step to ensure that the given dataset is adequate for multivariate modeling. At this point, key objectives and deliverables for the data audit were established and documented in a collaborative manner.
3. Multivariate Data Analysis
ProSensus worked closely with the client’s process experts to complete a multivariate data analysis of the entire sequential manufacturing process. ProSensus developed a global multi-block multivariate model that represented the complex manufacturing operation (shown below) from start to end.
This model enabled ProSensus to easily identify correlations between raw materials, different processing steps, and final quality. Incorporating all the blocks in a single multi-block multivariate model allowed ProSensus to simultaneously manipulate and optimize all the different process units (upstream and downstream units) in the manufacturing process. The model was also used to direct process improvement efforts to the most relevant process units. Clearly, such a model provides a powerful framework for plant-wide quality management.
Additionally, ProSensus was able to provide the client with actionable insights. Specifically, this data audit revealed 80% of the variation in product yield was explained by variation in raw materials (as illustrated in the observed vs. predicted plot on the left). This key finding enabled the client to set more stringent raw material specifications for their suppliers to improve the yield of the final product.
In addition to a plant-wide analysis, a data audit can also focus in on one or two process units. ProSensus developed a model solely on the furnace dataset for this manufacturer and was able to identify campaign to campaign differences impacting final product quality. As evident in the score plot below, there was an obvious distinction between the two production campaigns as seen in the clustering of the data points. The raw data plot below clearly shows that the furnace energy consumption across campaigns was significantly different. Although this seems like a simple finding, process experts are often without the time and/or resources to examine their historical data effectively. Focused data audits can often be of significant value.
4. On-Site & In-Hand Results
Finally, ProSensus holds an on-site presentation to review the key results and recommendations for future phases of work. ProSensus also delivers a comprehensive written report detailing the current status of data collection, results of multivariate analyses, and any future recommendations and areas of improvement.