Today’s businesses collect enormous amounts of data, whether they want to or not. Information about customers, suppliers, manufacturing processes, transportation and shipping, and computer and telecom networks builds up in the databases of nearly all businesses. Most companies don’t use much of it; they believe they have neither the time nor the expertise to extract meaningful intelligence from it.
But data analytics have changed the way many forward-thinking companies do business. While the raw data may be too much for humans to comprehend, an analytics solution can decipher the information and turn it into actionable intelligence that can improve operations in many ways.
Once assumed to be a tool among the computer networking realm, data analytics are being valued by many manufacturers, particularly in the area of equipment downtime. Unplanned downtime can cost manufacturers money on many fronts. For starters, there’s lost output and repair costs. But there is also wasted manpower (as employees idle) and unplanned overtime as production lines must later race to catch up. Marshall Institute, an asset consultancy company, has estimated that unplanned downtime costs a business three times as much as proper preventive maintenance.
While most manufacturers keep some type of log on their machines — such as the times they are serviced and by whom, how long they run, their output, and when they required human intervention -- it’s seldom in-depth enough to draw specific conclusions and make improvements in order to prevent unplanned downtime. While most equipment offers subtle clues in the lead-up to their failure, these signals are often missed on a busy production floor.
The idea of machine-to-machine (M2M) networking -- or embedding all equipment with smart sensors to communicate with each other, be monitored by an analytics solution, and be integrated with the company’s enterprise resource planning (ERP) or business process or asset management solution -- is a compelling one.
Working together, these solutions gather machine performance and process data of every kind, including age of equipment and components, last maintenance check, temperature, vibration levels, length of usage or cycle times, operations noises, oil temperatures and levels, the number of revolutions in each gear, and more. They then use the data in conjunction with historical information to accurately predict breakdowns in advance. Essentially, the machinery becomes “smart” and can inform workers or managers when attention or maintenance is required, often via a simple alert on a smartphone or tablet, hourly or daily reports, and dashboards on software interfaces.
Ultimately, the goal is to replace the usual cycle of “Break. Repair. Repeat.” Proactive, real-time preventive equipment maintenance is an approach that can ensure more efficient operations and improved profit margins. The industry uses the term “predictive maintenance” to underscore that technology helps determine precisely when a piece of equipment will require maintenance and eliminating the need for random preventive maintenance that may not be necessary. The technology could result in self-healing machinery that can schedule its own maintenance at the right times.
As with many advanced technologies, the idea of M2M networking and data analysis has trickled down from the aerospace and healthcare industries. GE’s jet engine division uses mounted sensors to monitor engine and component performance and ensure optimum operation. They can literally inform technicians exactly what maintenance needs to be done and when, potentially saving airlines hundreds of millions of dollars in fuel costs by keeping operating efficiency at the highest possible level. GE offers the same technology in its line of medical imaging equipment.
While machine failure on a shop floor might not have the kind of dire consequences of a failed jet engine, it can cut significantly into a company’s profits. While engineers and maintenance workers might “know” their machines very well through experience and intuition, they may be missing clues in how the machines operate in relation with other machines or the entire facility.
Eric Brethenoux, IBM’s director of predictive analytics, recently relayed to the web publication Data Informed a telling example of how machine-to-machine sensors, coupled with Big Data analytics, can uncover the root causes of large problems that humans miss.
German automaker BMW operates two identical engine cylinder production lines, one in Stuttgart and one in Munich. It was seeing important differences in the size of the cylinders being produced. Believing the differences were emanating from the styles of machine maintenance at the two locations, the company investigated this theory but ultimately couldn’t explain the discrepancies. What it did know was that the changes happened toward the end of shifts; the cylinders produced early in the shifts were consistent but would begin to deviate later.
Thanks to predictive analytics, BMW found that the bug wasn’t in the machinery; sunlight coming through a window in one facility was slowly heating up machinery to temperatures beyond optimum range and affecting the components being produced. Typically, a great deal of manpower is often wasted in fruitlessly searching for causes of mysterious machinery and operations effects.
Aside from preventing breakdowns and ensuring product quality, M2M technology, coupled with predictive analytics, can help companies optimize their spare parts warehousing and avoid unnecessary overstocking, reduce maintenance staff by eliminating unnecessary maintenance tasks, perform root-cause analysis of previous failures, and predict when future breakdowns are most likely to occur. It also allows manufacturers to extend the life of its equipment and get the most out of warranties.
In addition, predictive analytics are increasingly being used by equipment manufacturers to help them better understand when their customers are likely to make warranty claims. If faults in certain components are identified, manufacturers can take steps to replace parts in their customers’ equipment in advance and prevent expensive warranty claims or returns. Some M2M analytics and predictive maintenance solutions can be coupled with networked internal cameras, allowing operators or maintenance workers to literally see, via a tablet computer, inside the machines while they are working and check operations visually. Other sensors can “listen” to machine operations and hear anomalies that are not audible to human ears.
The opportunities being offered to manufacturers when it comes to Big Data and predictive analytics are extremely compelling. Companies can use them to take charge of their equipment and operations instead of the other way around.
