Caring for Your Mission-Critical Taskforce: Your Machines

A problem worth solving

When it comes to machines, there’s never really a good time for them to breakdown. Machinery breakdowns are inevitably costly. This is true for manufacturers of equipment, for their customers and also for providers of managed machine maintenance services. In response to the challenge, companies have explored a variety of strategies – even preventive maintenance programs - to increase machine uptime. The goals have always been: ensure high availability of machines, rationalize service and repair costs across the network, and incorporate learning into the manufacturing and application of new models of machines. But every strategy – even a combination of strategies - has had only limited success.

Finding and framing the real problem

Because preventive maintenance relies on average or expected machine-life statistics, to predict when maintenance will be required, it often fails to deliver expected machine uptime results. The real challenge then is to determine the actual condition of each individual in-service equipment in order to predict when maintenance should be performed. This approach promises cost savings over routine or time-based preventive maintenance, because tasks are performed only when warranted. The key then is to access insights into the state-of-the-machine at the right time. By knowing precisely which equipment needs maintenance, machine upkeep can be better planned (spare parts, people, etc.) and what would have been ‘unplanned stops’ can be transformed to shorter and fewer ‘planned stops’, thus increasing availability.

Solving the Problem

Juxtaposition data from machines across multiple sites, over months of usage, with near real-time data from individual machines that must be maintained, to derive insights that provide warning signals and alert managers to potential machine failure. Here’s how it works:

Leverage our insights-as-a-service offering to curate information from machines like the one that must be maintained - from across a wider landscape. For example, we curated four million failure ticket records from over 8,500 ATMs to develop, train and test a machine learning model for predicting ATM failure in North America. Harvest diverse data – past events of dysfunction, maintenance schedules, log data, transaction load, time since last repair, age of machine, and defects reported.

Ingest the data into an Apache Spark data processing engine for logistic regression. The Spark open source cluster computing framework cleanses and enriches data as quickly as in 27 seconds. In the case of the ATM predictive maintenance program, the logistic regression algorithms predicted ATM fault in 60 milliseconds with 80% accuracy.

Use Tableau visualization tools to present reports for interpretation and analysis. The color-coded dashboard helps maintenance teams review notifications, service calls and failure patterns over time periods, cities / states, and model types.

The Outcomes

  • Service-level guarantees with up to 99% machine availability
  • Up to 18% reduction in costs of unwarranted preventive maintenance and repair
  • Deep optimization of technician allocations and dramatic improvements in the productivity of maintenance engineers. In the context of ATM maintenance management, the solution delivered a 14.3% increase in operational efficiency and reduced daily average service calls per technician to 4 from 3.
  • Insights derived from an exhaustive study of maintenance logs can help refine the design of the next generation of the machine. In the context of the ATMs, insights revealed the propensity for the mechanical motor of the ATM to fail faster than the printer or the keypad.