Jan 29, 2016

Prognosis is good for the manufacturing industry

Reactive maintenance: It is no longer enough
Estimates indicate that reactive maintenance of equipment within the manufacturing industry accounts for nearly $200¹ billion in annual costs in the US alone. In addition, operations in the heavy equipment-based manufacturing sector may have up to 60% of their total spend locked-up in reactive maintenance. These figures suggest that the maintenance cycle conversion, which tracks the transition from reactive to planned maintenance, will continue to be a critical focus as organizations strive to achieve end-to-end operational excellence.

To that end, manufacturers and operators alike are adopting predictive maintenance strategies - what we call 'prognostics'. They are using machine learning techniques and high-fidelity sensing equipment to analyze real-time condition data from their most expensive and mission-critical assets.

What does prognostics entail?
The global market for sensors is estimated to be nearly $76BN by 20172. It has been estimated that the total number of things that are "connectable" in an IoT world will reach 212 billion by 20203. Of these, at least 20-30 billion will actually be connected by 2020.

Some estimates have pegged the total amount of current human and machine data at nearly 2.7 Zettabytes and growing.5 The current trend points to a machine-data heavy future given the drive towards a fully connected future coupled with a growing ubiquity of industrial automation.

On a more practical note, consider the example of the amount of data generated by a typical sensor output from a typical jet engine:

"Boeing jet engines can produce 10 terabytes of operational information for every 30 minutes they turn. A four-engine jumbo jet can create 640 terabytes of data on just one Atlantic crossing, multiply that by the more than 25,000 flights flown each day."4

We believe that to be relevant and useful, there are essentially three big challenges that must be addressed holistically by means of an end-to-end solution:

  • Data integration
  • Data analytics
  • Insight-to-Action

Going beyond prediction
While predicting the failure of parts well in advance has innumerable benefits, the real differentiator is understanding how a failure manifests itself, and providing actionable recommendations that empower engineering and maintenance crews to improve the remaining useful life of the asset in service. The process this approach describes has been dubbed "prognosis," and combines elements of diagnostics, predictive analytics, and prescriptive performance engineering to identify highly functional solutions that mitigate mission-critical incidents.

Solutions like Genpact's Digital EngineerTM play a key role as the centerpiece of a detailed transformational strategy that leverages the industrial internet of things (IIoT) to drive process outcomes.

Over the longer term, the system becomes a virtual expert at offering the highest quality recommendations and solutions to operations teams. In the event that no mitigating solution is possible, the system thinks through the logistics of corrective maintenance to ensure the availability of the right spare parts. It engages the entire supply chain to ensure the availability of safety stocks, and sends an approval request through all the appropriate channels to acquire new stock that compensates for the depletion. This can be particularly valuable in scenarios where a failure impacts a critical system that is not only costly to replace, but may also be difficult to acquire due to long cycle times for additional parts.

The net effect of combining human and machine intelligence on process operations is the ability to achieve continuous process optimization that drives incident rates and corresponding downtime to zero.

Author: Pradyumna S. Upadrashta, Ph.D. - Chief Science Officer, Analytics & Research

  1. Introduction to Predictive Maintenance, 2nd edition, written by R. Keith Mobley 
  2. PRWEB
  3. EMC
  4. Information Management
  5. Dan Vesset, Benjamin Woo, Worldwide Big Data Technology and Services 2012– 2015, IDC March 2012