IoT: Bringing Big Data to Manufacturing
Creating a competitive advantage is hard work. It sometimes feels like all the innovations have already hit the market and the margin between you and the competition is growing narrower. The modern marketplace requires vast amounts of information to be collected and analyzed beyond what engineers and marketers can eyeball. And from that, the advantage we seek is a single key insight culled from a big haystack of data.
Big data is one of the popular buzzwords that describes a vast set of information, something bigger than what can be analyzed through traditional (read: human) means. It starts with millions of records, which can seemingly be unrelated to reach other. You can’t just throw this data set into Excel and make a Pareto chart. Storage requirements alone are daunting. Gigabytes could quickly turn into terabytes of information. Additionally, organizations need to invest in Information Technology infrastructure with hardware and manpower to maintain the data. The range is an endless ocean of 0’s and 1’s, for which algorithms and models are needed to identify the patterns and trends that ultimately generate value, and offset the cost of managing this raw information. It seems like a game rigged for only the largest of corporations, and Internet of Things (IoT) is in process of making all manufactures required to play.
Historical PLM Data
Product Lifecycle Management (PLM) has always been described as complete virtual lifecycle of a product: from cradle to grave. But once the product is in the customers’ hands all bets are off – the opportunity to intimately understand how the consumer uses the product is lost. In a previous article I expanded on how PLM was a system for continuous improvement, however there is a deficit of information on what to improve post-release. The only lifecycle information received post-production are often complaints. These may translate to actionable changes through internal Correction Action Requests (CAR), but this isn’t proactive and it has a long cycle time. Fortunately, advancements in network communications, analysis tools and cost-effective sensors have brought about a new buzzword – Internet of Things, which describes connected things; some of which are connected devices sending sensor data to manufactures through an Internet connection.
There is more to Internet of Things than connected devices capturing and relaying information. It can also be human resources, such as a field technician receiving maintenance information from a connected thing. It can also be Information Technology servers automating the process of sending a technician out to the field to make a repair. Or, it could be a professional athlete sending performance data back to his coaching staff for intense scrutiny and analysis.
IoT: Dawn of Big Data
However IoT innovations are applied across different industries, the byproduct is the same: a lot of data. As previously stated, the influx of data requires a centralized location to store and maintain it. The PLM environment provides an optimal infrastructure for IoT data captured from products in the field when that environment is equipped with a fully defined engineering Bill of Material (eBOM) per product configuration and a serialized instance for every manufactured product. Put more simply, the virtual counterpart of every manufactured product is refered to as the digital twin. Think of it in terms of a car. While you’re driving around town your car is sending information back to the auto maker via sensors. That data is captured as a record that represents your physical vehicle within the PLM tool. That record (and, therefore, your car’s individual performance) can be analyzed along with every other digital twin to find insights on ways to improve product performance, therefore improving consumer loyalty and sales. Tools are available to virtualize every serialized product in the marketplace; and manufacturers better start taking advantage of this technology to stay competitive . Entire product histories, as captured from sensors, are mapped in the PLM environment. This history grows as long as products are in use by the customer. Now, it’s truly cradle to grave. But that’s a lot of information, and, without a meaningful way of optimizing it, a lot of new overhead.
IoT is being heavily positioned for serviceability. Analysis is conducted from sensor information across an entire product line or fleet, allowing for proactive service and maintenance. If the oil levels on your car are low, you will be automatically scheduled for a service appointment. If a particular widget is trending with excessive ware across all customers, a service campaign can be launched. Customer satisfaction and repeat business increases. A blueprint for sensor placement can be done by conducting a Failure Mode Effects Analysis (FMEA). FMEA provides a mapping of critical areas with predicted fault tolerances of a product including additional information on both failure types and conditions.
True value is generated when statistical analysis is conducted across IoT data and other business information. To manage this type of information, we can look to machine learning. Machine learning provides the business intelligence to make smart impactful decisions through the identification of patterns. We can turn unknown unknowns into knowns. Impact analysis can go deeper than resource and material costing. Greater efficiencies of project resources management are found through aligning human capital with the optimal skill set and resource load. Smart metrics allow team members to make better decisions, and management to steer the business in a better direction.
Where we are going
IoT will bring product Big Data to manufacturing companies allowing exhaustive analysis across multiple, and seemingly unconnected, stacks of data. For example, companies can improve or introduce new products after comparing device usage against customer behavior patterns discovered from marketing studies. Think of the power of targeted marketing you see on Facebook and Google Adwords, and applying that precision to manufactured goods. Targeted marketing tracks usage patterns and serves advertising for products that match what individuals are most likely to consume. With advancements in production output like 3D printing, manufacturing cycle time will continue to shrink. Companies are already creating products using additive manufacturing and 3D printing to get them to market faster. Manufactures, seemingly instantly, will be releasing iterations driven from patterns in customer behavior. Cost effective material changes will be conducted through better forecasting of raw material markets. The concept of IoT goes beyond just smart connected devices. It’s really about harnessing big data and transforming businesses into agile and efficient juggernauts.
