By Chris Penrose, COO, FogHorn
With 76% of consumers experimenting with new shopping behavior during the COVID-19 pandemic, consumer goods organizations are being challenged to control operations as tightly as possible to soften the impacts of sudden changes in demand. However, due to a lack of real-time insight, consumer packaged goods (CPG) companies often struggle with minimizing scrap rates and improving product yield — both of which are critical to maintaining operational margins and assuring satisfied customers as demand scales over time.
To address this challenge, CPG organizations are seeking out edge-based, AI-driven, real-time streaming analytics technology to provide increased visibility into production line performance. By doing so, organizations will be able to leverage valuable insights from manufacturing data as it is produced and more closely and accurately control manufacturing operations.
A typical CPG scenario
To understand the value edge-based, real-time streaming capabilities deliver to CPG organizations, consider the following scenario: A CPG company has a facility that produces single-use containers, fills each container with product, and then packages multiple containers together in a larger package to be sold in retail outlets. The company employs an automated packaging system and programmed logic controller (PLC) that are designed to fill each container based on weight. When raw product is dispensed from the tank into each container, this packaging function must account for all these variables with a simple measurement, while also considering how much of the product has been used.
As part of their regular operations, the organization pulls product weight samples 1-2 times every hour or so from their data historian. Volume settings are frequently adjusted to ensure the proper weight, and the final packaged product would also be sampled for accuracy. When the company catches any errors in weight during sampling, the product is scrapped, hurting margins. Moreover, when they don’t catch these errors, the out-of-spec product is sent to market – potentially impacting brand reputation.
Like many CPG organizations, this company is hesitant to reprogram PLC logic, as changes to the PLC are often disruptive to productivity and can have detrimental effects on operations. This company is also sensitive to streaming sensitive product data to a third-party cloud because of stringent security and governance requirements.
The benefit of edge-based, AI-driven, real-time streaming analytics
By adopting edge-based, AI-driven, real-time streaming analytics, this organization gains major new insights to overcome these challenges and ultimately reduce scrap rates and boost product yield.
With a robust edge analytics solution in place, this company’s packaging system can be expanded to run scales in parallel at a much greater magnitude, increasing the volume of sampled data as well as the frequency of sampling. The scale measures can then be fed into the analytics system, where AI models instantly deliver visibility showing how packages are doing on a statistical basis.
As a result, the organization gains comprehensive insights that make it possible to identify most defective products while they are still on the production line, rather than after production is complete. Moreover, with this visibility, far fewer defective products would leave the facility and go to market.
Using AI, the organization does not have to make any changes to PLC logic, ensuring work on the factory floor continues uninterrupted. And because all processing is done at the edge, with compute-intensive calculations executed entirely on the factory floor, the company can avoid having sensitive data leave the facility, eliminating potential security risks associated with sending data to the cloud and maintaining any data governance requirements.
This hypothetical scenario highlights just a few basic ways that CPG companies can use AI-based streaming analytics at the edge to improve product yield and scrap rates. However, there are many other ways that such manufacturers could employ this technology on their vast quantities of sensor data to boost operational efficiency. These include many important applications for streaming video data. Aside from catching defects, streaming analytics on video data can determine if an excess of material is being used to manufacture a product, helping avoid waste and lower costs. In addition, it can be used to identify any potential safety hazards that may be present on a factory floor, protecting workers and averting accidents that could disrupt production.
Edge-based, AI-driven streaming analytics supports many other use cases involving sensor data beyond streaming video. For one, data entry can be automated, eliminating thousands of hours that workers might normally spend to manually log data in any given year. It can also improve the health of machinery via predictive maintenance and uncover opportunities to reduce energy consumption.
Increase competition with edge computing
CPG manufacturing organizations must find ways to improve product yield and reduce scrap rates while addressing sudden shifts in consumer demands, such as those that occurred at the outset of the pandemic. With consumer purchasing habits likely to continue fluctuating rapidly in the foreseeable future, CPG companies should look to edge-based, AI-enabled, real-time streaming analytics technology to give them the flexibility and insights to stay ahead in an increasingly competitive global marketplace.