Historically, maintenance has been driven by a preventative schedule. Today, preventative maintenance, where actions are performed regardless of actual condition, is giving way to Predictive, or Condition-Based, maintenance, where actions are based on actual, real-time insights into operating conditions. While both are far superior to traditional Corrective maintenance (action only after a piece of equipment fails), Predictive is by far the most effective. In fact, McKinsey points to a 50% reduction in downtime and a 40% reduction in maintenance costs when using IoT and data analytics to predict and prevent breakdowns.
Determining what specific activities need to be performed based on an asset’s actual condition or state, is all about predicting and preventing failures to avoid costly unplanned downtime or more costly repairs. The key is active and ongoing monitoring of prognostic health data. By capturing and analyzing this data, agencies can learn how external forces are affecting fleet operation, including everything from weather, terrain, and loading to operator actions such as hard acceleration or braking. I will talk more about Predictive Maintenance in the Public Sector and how government agencies are optimizing equipment maintenance with data insight during Cloudera’s Industry Event on August 5.
Predicting when equipment needs maintenance based on actual usage patterns and mitigating downtime drives operational efficiency, therefore Predictive maintenance can deliver both improved availability and safety.
Examples of predictive maintenance we’ve delivered for customers at Cloudera include:
Sikorsky, the helicopter division of Lockheed Martin, is uncovering patterns in aircraft performance and parts that improve flight safety, optimize aircraft operations, and significantly reduce costs.
To achieve its goal to be the safest OEM of helicopters, the company wanted to conduct more advanced analytics so it could better predict and resolve potential issues before they caused flight groundings or safety issues.
Using a scalable data management and analytics platform built on Cloudera Enterprise, Sikorsky can process and store data in a reliable way, and analyze full data sets across entire fleets. The use of Cloudera Enterprise enables staff to focus on aviation and business challenges instead of worrying about monitoring the cluster or platform upgrades.
Navistar plays a critical role in the public sector, as a major provider of commercial heavy trucks, vehicles and school buses. Here, reliability is critical as breakdowns are unacceptable with the potential to impact supply chains or strand students.
Navistar relies on predictive maintenance, which leverages IoT and data analytics to predict and prevent breakdowns of commercial trucks and school buses. “We use the Cloudera tool to employ machine learning for preventive maintenance,” says Terry Kline, Navistar SVP and CIO. “Cloudera is foundational in how we track and govern our data.”
Navistar, has developed an IoT-enabled diagnostics platform on Cloudera Enterprise. Massive amounts of data on engine performance, truck speed, acceleration, coolant temperature, and brakes are all continually analyzed. The result: Navistar has reduced maintenance costs up to 40% and seen an impressive 80% reduction in catastrophic breakdowns.
The same techniques can enable public agencies to more accurately predict performance and improve readiness of vehicles, power plants, ports, and other key assets. In addition, the US Department of Energy recently released a report concluding that an effective predictive maintenance program is 8 to 12 percent more cost effective than a program that relies solely on preventative maintenance.
Factors to be considered in when implementing a predictive maintenance solution:
- Complexity: Predictive maintenance platforms must enable real-time analytics on streaming data, ingesting, storing, and processing streaming data to instantly deliver insights.
- Data volume and variety: The platform must handle a wide variety of data types, from intermittent readings of sensor data (temperature, pressure, and vibrations) to unstructured data (e.g., images, video, text, spectral data) or other input such as thermographic or acoustic signals.
- Silos: Information from existing systems of record can be combined with new sources of structured and unstructured data to provide additional data in support of analytics and machine learning model development.
- Cost: Traditional data management tools are notoriously expensive, difficult to scale, and unsuited to processing petabytes of streaming data. Organizations need a secure, governed, scalable platform that can easily ingest, store, manage, and process streaming data—at a lower cost.
- Capability: Predictive modeling capabilities are key to delivering insights, and traditional platforms provide limited modeling or machine learning capabilities to predict or prevent anomalies or disturbances.
Effective predictive maintenance is more than just IoT sensors and algorithms driving improved uptime. It is the foundational realization that enterprise data is at the heart of any predictive maintenance initiative. Cloudera delivers this with an end-to-end platform enabling secure, governed data ingest, storage and analysis, from edge to AI.
More than 40 global governments rely on Cloudera to deliver data insight. To learn more about the great work Cloudera is doing in the Public Sector, please visit our government solutions page on the website.